Search

Major developments in the e-commerce business have an impact on supporting industries, such as the logistics industry. Logistics service providers for e-commerce consumers must focus on the quality of their services. Service quality is determined by meeting customer expectations, including on-time delivery, parcel security, and customer service. In the process of sending parcels, one thing that affects the timeliness of delivery is the last-mile delivery stage. Due to a large number of last-mile stations and the geographic dispersion of their locations throughout Indonesia, the service quality between stations becomes inhomogeneous. The performance of the last-mile station can be measured by comparing the relative efficiency between stations. In this study, the initial stage of the efficiency analysis of the last-mile delivery station is by conducting a cluster analysis that divides the station into two groups, namely the Leader and Majority clusters, which aim to group homogeneous DMUs. Efficiency analysis with DEA is given for each cluster. The result shows that 94 of the 133 stations are relatively efficient for the Leader cluster stations, and 136 of the 466 stations are relatively efficient for the Majority cluster stations. A decision tree method is used for classification modeling to determine the characteristics of a relatively efficient station. Total daily delivery becomes the initial variable that determines the station to enter the Leader or Majority cluster. The second criterion is the variable number and quantity of goods delivered on time.

Aquesta tesi consisteix a desenvolupar models d'aprenentatge automàtic per predir mètriques en l'àmbit del disseny de microxips. S'ha desenvolupat en col·laboració amb Marvell Technology, Inc., que és una empresa dedicada al disseny i producció de microxips. Els microxips es dissenyen i es posen a prova mitjançant alguns processos que requereixen gastar molts recursos. Aquests recursos s'han de sol·licitar prèviament i es poden malgastar si la sol·licitud no coincideix amb la realitat. Això pot costar a l'empresa molts diners i temps. L'objectiu principal d'aquest projecte és resoldre aquest problema mitjançant algorismes i tècniques d'aprenentatge automàtic. Es realitza la creació de diferents models per trobar la solució més òptima a aquest problema. Aquests models d'aprenentatge automàtic s'alimenten de dades obtingudes tant de projectes experimentals com reals de l'empresa. Les dades tenen molts problemes com ara inconsistències i valors atípics. Per tant, es fa un preprocessament adequat, així com algunes transformacions per obtenir informació valuosa per als models. El resultat d'aquest projecte és la troballa d'un model que utilitza l'aprenentatge automàtic per poder predir aquestes mètriques amb èxit per reduir en gran mesura el malbaratament de recursos., This thesis consists in developing machine learning models to predict metrics in the field of microchip design. It has been developed in collaboration with Marvell Technology, Inc., which is a company dedicated to the design and production of microchips. Microchips are designed and tested through some processes that require spending many resources. These resources need to be requested beforehand, and they can be wasted if the request doesn't match reality. This can cost the company a lot of money and time. The main goal of this project is to solve this problem using machine learning algorithms and techniques. The creation of different models is carried out to find the most optimal solution for this problem. These machine learning models are fed with data obtained from both experimental and real projects of the company. Data comes with many problems like inconsistency and outliers. Therefore, proper preprocessing is done, as well as some transformations to get valuable information for the models. The outcome of this project is the finding of a model that uses machine learning to be able to predict these metrics successfully to strongly reduce the waste of resources.

Malware is any intrusive program that exploits system vulnerabilities on a computing system. A malware infection can be disastrous for any organization. It can cripple networks and systems, as well as destroy, delete, corrupt, or exfiltrate sensitive data. In this thesis we explore how machine learning can detect and classify malware threats to prevent further damage in a network. Specifically, we use Naive Bayes, K-Nearest Neighbours, Decision Tree, Random Forest and Support Vector Machine algorithms, determining the precision, accuracy and recall averages for 100 different random states, as well as the variance of each one of these averages. We will use two datasets: the CIC-Evasive-PDFMal2022 dataset (which consists of 10025 records and 28 numerical features) and the CIC-MalMem-2022 dataset (which consists of 58596 records and 55 numerical features). For malware classification, we will only use the second one because the first one is not labeled with the malware type.

This paper presents a study on forecasting the stock close price of ACB bank from 2012 to 2022 using various learning machine models. The models used in this study include Decision Tree, Random Forest, and LSTM, which are combined with sentiment analysis for Vietnamese news using the Pho Bert approach. To evaluate the performance of the models, R2 and RMSE are employed as evaluation metrics. The results indicate that the LSTM model with news sentiment analysis provides the best performance in both evaluation metrics. This study contributes to the understanding of the effectiveness of combining machine learning models with sentiment analysis for forecasting stock prices.

Abweichender Titel laut Übersetzung der Verfasserin/des Verfassers, Dissertation Universität Innsbruck 2023

The global tech company Ericsson has been tracking the return rate of their products for over 30 years, using it as a key performance indicator (KPI). These KPIs play a critical role in making sound business decisions, identifying areas for improvement, and planning. To enhance the customer experience, the company highly values the ability to predict the number of returns in advance each month. However, predicting returns is a complex problem affected by multiple factors that determine when radios are returned. Analysts at the company have observed indications of a potential correlation between alarm data and the number of returns. This paper aims to address the need for better prediction models to improve return rate forecasting for radios, utilizing alarm data. The alarm data, which is stored in an internal database, includes logs of activated alarms at various sites, along with technical and logistical information about the products, as well as the historical records of returns. The problem is approached as a classification task, where radios are classified as either "return" or "no return" for a specific month, using the alarm dataset as input. However, due to the significantly smaller number of returned radios compared to the distributed ones, the dataset suffers from a heavy class imbalance. The imbalance class problem has garnered considerable attention in the field of machine learning in recent years, as traditional classification models struggle to identify patterns in the minority class of imbalanced datasets. Therefore, a specific method that addresses the imbalanced class problem was required to construct an effective prediction model for returns. Therefore, this paper has adopted a systematic approach inspired by similar problems. It applies the feature selection methods LASSO and Boruta, along with the resampling technique SMOTE, and evaluates various classifiers including the Support vector machine (SVM), Random Forest classifier (RFC), Decision tree

Padi adalah salah satu bahan pangan yang menjadi sumber utama bagi masyrakat indonesia. Padi bukan hanya sebagai sumber bahan makanan namun juga menjadi ladang untuk mencari nafkah bagi kaum petani. Namun sering kali kita menemui kasus dimana sering terjadinya gagal panen dikarnakan wabah atau penyakit, baik itu di akibatkan oleh hama ataupun kondisi iklim yang tidak mendukung. Penyakit pada padi sendiri ada cukup banyak dan masing masing dari penyakit tersebut memiliki penanggulangan atau pencegahan yang berbeda pula. Maka dari itu dibuatlah sebuah sistem yang dapat mengklasifikasikan jenis penyakit yang terdapat pada tanaman padi. Penelitian ini mengusulkan Klasifikasi Penyakit Pada Padi dengan Ekstraksi Fitur LBP dan GLCM. Dalam penelitian ini digunakan 120 dataset citra daun padi. Proses dan metode yang diusulkan yaitu Akuisisi citra, preprocessing, segmentasi dengan metode konfersi nilai menjadi biner, operasi morfologi dengan menggunakan Area Open, ekstraksi fitur menggunakan LBP dan GLCM, dan terakhir tahap klasifikasi dengan Decision Tree. Berdasarkan hasil pengujian, diperoleh tingkat akurasi sebesar 90%. Diharapkan dengan adanya sistem yang dibuat ini dapat membantu sebagian besar para petani untuk mencegah terjadinya penyakit yang terjangkit pada padi, sehinggak hasil panen diharapkan dapat lebih meningkat dengan kualitas yang baik.

Padi adalah salah satu bahan pangan yang menjadi sumber utama bagi masyrakat indonesia. Padi bukan hanya sebagai sumber bahan makanan namun juga menjadi ladang untuk mencari nafkah bagi kaum petani. Namun sering kali kita menemui kasus dimana sering terjadinya gagal panen dikarnakan wabah atau penyakit, baik itu di akibatkan oleh hama ataupun kondisi iklim yang tidak mendukung. Penyakit pada padi sendiri ada cukup banyak dan masing masing dari penyakit tersebut memiliki penanggulangan atau pencegahan yang berbeda pula. Maka dari itu dibuatlah sebuah sistem yang dapat mengklasifikasikan jenis penyakit yang terdapat pada tanaman padi. Penelitian ini mengusulkan Klasifikasi Penyakit Pada Padi dengan Ekstraksi Fitur LBP dan GLCM. Dalam penelitian ini digunakan 120 dataset citra daun padi. Proses dan metode yang diusulkan yaitu Akuisisi citra, preprocessing, segmentasi dengan metode konfersi nilai menjadi biner, operasi morfologi dengan menggunakan Area Open, ekstraksi fitur menggunakan LBP dan GLCM, dan terakhir tahap klasifikasi dengan Decision Tree. Berdasarkan hasil pengujian, diperoleh tingkat akurasi sebesar 90%. Diharapkan dengan adanya sistem yang dibuat ini dapat membantu sebagian besar para petani untuk mencegah terjadinya penyakit yang terjangkit pada padi, sehinggak hasil panen diharapkan dapat lebih meningkat dengan kualitas yang baik.

Account receivables (AR) are a business's most valuable asset. This research paper explores how supervised machine learning can predict payment outcomes for open invoices, specifically addressing the common challenge of maintaining consistent income for small to medium enterprises (SMEs). We developed a model using various machine learning techniques, including linear regression, random forest regressor, decision tree classifier, Linear SVR, and XGB Regressor, by training it on actual AR data. This model assists collectors in forecasting debt payment.

HR analytics is a leveraging technology in which the process of HR business is automated and to develop data driven insights to inform talent decisions. In this project, the main objective is to eliminate or to reduce the HRs job and make the selection process automated. In order to achieve this, the project was divided into two modules. One module processes the resumes and parses them to extract the required data. Firstly, the resume parsing is done using the Named entity recognition (NER) method using the spacy model. The resumes are annotated using NER annotator which labels the required text/ data in the resumes. The second module wants to make the candidate hiring predictions automated. The candidates will be checked on different aspects like experience, their stand towards one company i.e. stability which tells how stable the candidates are in staying with one company. The tenure with the current company they are working for, which tells how many months the candidate is working in the current company when they are applying for the job. The first module gave results that were accurately obtained after using the Named entity recognition method to parse the resumes and extract the skills from the resumes. After successfully extracting the correct skills from resumes with high accuracy, it was matched with the Job ID which matches the skills with the required data in the job description. The second module of the project gave the results based on a decision tree which was plotted and made decisions whether the candidate profile is fitted or not. The prediction is based on training the decision tree ML model which gave an accuracy of 70 %. The precision, recall and f1 score for the data is obtained and checked how well the data is trained. Overall project will make the HR job easier and automate the process which will help to reduce the complete time of recruitment.

Women’s football is currently a phenomenon in expansion, both in the number of practitioners and federative records. Scientific research must be at the forefront of this growth, proposing solutions with empirical support that help improve performance not only during training but also during competition. The objective of this study was to propose a multivariate model based on the decision tree technique, with the aim of classifying and predicting the criteria that modulated the attack depending on the part of the match (either the first or second part). To do this, 6063 attacks from the two most recent world championships have been collected and analysed. The available results indicate that, although more attacks occur during the first part, it is in the second part when the goals are scored. In addition, the presented model confirms that the most successful attacks (goal, shot or sending to the area) occur with the score winning or losing, and with a duration of less than 20 seconds. The results of this study can help female soccer coaches to improve the training process of offensive actions and attack

Objective Using supervised machine learning to classify the severity of cardiovascular autonomic neuropathy (CAN). The aims were 1) to investigate which features contribute to characterising CAN 2) to generate an ensembled set of features that best describes the variation in CAN classification. Methods Eighty-two features from demographic, beat-to-beat, biochemical, and inflammation were obtained from 204 people with diabetes and used in three machine-learning-classifiers, these are: support vector machine, decision tree, and random forest. All data were ensembled using a weighted mean of the features from each classifier. Results The 10 most important features derived from the domains: Beat-to-beat, inflammation markers, disease-duration, and age. Conclusions Beat-to-beat measures associate with CAN as diagnosis is mainly based on cardiac reflex responses, disease-duration and age are also related to CAN development throughout disease progression. The inflammation markers may reflect the underlying disease process, and therefore, new treatment modalities targeting systemic low-grade inflammation should potentially be tested to prevent the development of CAN. Significance Cardiac reflex responses should be monitored closely to diagnose and classify severity levels of CAN accurately. Standard clinical biochemical analytes, such as glycaemic level, lipidic level, or kidney function were not included in the ten most important features. Beat-to-beat measures accounted for approximately 60% of the features in the ensembled data., Objective: Using supervised machine learning to classify the severity of cardiovascular autonomic neuropathy (CAN). The aims were 1) to investigate which features contribute to characterising CAN 2) to generate an ensembled set of features that best describes the variation in CAN classification. Methods: Eighty-two features from demographic, beat-to-beat, biochemical, and inflammation were obtained from 204 people with diabetes and used in three machine-learning-classifiers, these are: support vector machine, decision tree, and random forest. All data were ensembled using a weighted mean of the features from each classifier. Results: The 10 most important features derived from the domains: Beat-to-beat, inflammation markers, disease-duration, and age. Conclusions: Beat-to-beat measures associate with CAN as diagnosis is mainly based on cardiac reflex responses, disease-duration and age are also related to CAN development throughout disease progression. The inflammation markers may reflect the underlying disease process, and therefore, new treatment modalities targeting systemic low-grade inflammation should potentially be tested to prevent the development of CAN. Significance: Cardiac reflex responses should be monitored closely to diagnose and classify severity levels of CAN accurately. Standard clinical biochemical analytes, such as glycaemic level, lipidic level, or kidney function were not included in the ten most important features. Beat-to-beat measures accounted for approximately 60% of the features in the ensembled data.

Due to significant technology advancements, people's needs have expanded. As a result, there have been more requests for loan approval in the banking sector. A few qualities, taken for consideration, when choosing a candidate for loan approval in order to, determine loan's status. Banks face a major challenge; when it, comes to assessing loan applications and lowering the risks associated with potential borrower defaults. Since they must thoroughly evaluate each borrower's eligibility for a loan, banks find this process to be particularly challenging. This research proposes combining machine learning (ML) models and ensemble learning approaches to find the probability of accepting individual loan requests. This tactic can increase the accuracy with which qualified candidates are selected from a pool of applicants. As a result, this method can be used to address the problems with loan approval processes outlined above. Both the loan applicants and the bank employees profit from the strategy's dramatic reduction in sanctioning time. Because of the banking industry's expansion, more people were applying to loans at banks. In order to predict the accuracy of loan approval status for applied person, we used four different algorithms namely Random Forest, Naive Bayes, Decision Tree, and KNN. By using these, we obtained better accuracy of 83.73% with Naïve Bayes algorithm as best one.

A very challenging and difficult part of finding a cure for Rheumatoid Arthritis ( RA) disease is the same as finding a cure for an autoimmune disease. In the case of autoimmune diseases, it is very difficult to find the cause. RA is also one of the autoimmune diseases that causes inflammation in joints. It does not only limit to joints but also can spread or occur in various parts of the body like skin, lungs, and many others. Many features like age, sex, and previous history of certain diseases affect the occurrence of RA. The proposed ML model uses a decision tree algorithm to analyze a comprehensive set of clinical and laboratory variables collected from RA patients. These variables include age, ID, time, treatment, gender, and other baseline features. The ML model employs feature selection techniques to identify the most relevant predictors that contribute to disease progression. By training the decision tree algorithm on a large dataset of RA patients, the model generates an accurate predictive model for assessing disease severity and progression. The model is able to detect RA disease with almost 80% accuracy with the given training dataset. RA can have many other symptoms and characteristics. This data was trained on a basic dataset available from the Kaggle website. The dataset is in CSV format and has 6 features. The model was trained using a decision tree algorithm, which categorizes the data into different categories and then checks for specific data when needed in Jupyter Notebook. The study explored detecting basic symptoms and characteristics of RA disease. The Decision tree model has given accuracy of 86.5% and 74.4% on training and testing respectively. So Accuracy of the model is pretty good for the dataset used.

Major developments in the e-commerce business have an impact on supporting industries, such as the logistics industry. Logistics service providers for e-commerce consumers must focus on the quality of their services. Service quality is determined by meeting customer expectations, including on-time delivery, parcel security, and customer service. In the process of sending parcels, one thing that affects the timeliness of delivery is the last-mile delivery stage. Due to a large number of last-mile stations and the geographic dispersion of their locations throughout Indonesia, the service quality between stations becomes inhomogeneous. The performance of the last-mile station can be measured by comparing the relative efficiency between stations. In this study, the initial stage of the efficiency analysis of the last-mile delivery station is by conducting a cluster analysis that divides the station into two groups, namely the Leader and Majority clusters, which aim to group homogeneous DMUs. Efficiency analysis with DEA is given for each cluster. The result shows that 94 of the 133 stations are relatively efficient for the Leader cluster stations, and 136 of the 466 stations are relatively efficient for the Majority cluster stations. A decision tree method is used for classification modeling to determine the characteristics of a relatively efficient station. Total daily delivery becomes the initial variable that determines the station to enter the Leader or Majority cluster. The second criterion is the variable number and quantity of goods delivered on time.

This paper presents a study on forecasting the stock close price of ACB bank from 2012 to 2022 using various learning machine models. The models used in this study include Decision Tree, Random Forest, and LSTM, which are combined with sentiment analysis for Vietnamese news using the Pho Bert approach. To evaluate the performance of the models, R2 and RMSE are employed as evaluation metrics. The results indicate that the LSTM model with news sentiment analysis provides the best performance in both evaluation metrics. This study contributes to the understanding of the effectiveness of combining machine learning models with sentiment analysis for forecasting stock prices.

Weed resistance to control measures, particularly herbicides, is a growing problem in agriculture. In the case of herbicides, resistance is sometimes connected to genetic changes that directly affect the target site of the herbicide. Other cases are less straightforward where resistance arises without such a clear-cut mechanism. Understanding the genetic and gene regulatory mechanisms that may lead to the rapid evolution of resistance in weedy species is critical to securing our food supply. To study this phenomenon, we exposed young Arabidopsis plants to sublethal levels of one of four weed management stressors, glyphosate herbicide, trifloxysulfuron herbicide, mechanical clipping, and shading. To evaluate responses to these stressors we collected data on gene expression and regulation via epigenetic modification (methylation) and small RNA (sRNA). For all of the treatments except shade, the stress was limited in duration, and the plants were allowed to recover until flowering, to identify changes that persist to reproduction. At flowering, DNA for methylation bisulfite sequencing, RNA, and sRNA were extracted from newly formed rosette leaf tissue. Analyzing the individual datasets revealed many differential responses when compared to the untreated control for gene expression, methylation, and sRNA expression. All three measures showed increases in differential abundance that were unique to each stressor, with very little overlap between stressors. Herbicide treatments tended to exhibit the largest number of significant differential responses, with glyphosate treatment most often associated with the greatest differences and contributing to overlap. To evaluate how large datasets from methylation, gene expression, and sRNA analyses could be connected and mined to link regulatory information with changes in gene expression, the information from each dataset and for each gene was united in a single large matrix and mined with classification algorithms. Although our mo

Overcooked, an immersive multiplayer video game centered around cooperative cooking challenges, provides the roots for this research project. The study focuses on designing and evaluating a hand-authored controller in comparison to controllers implemented using various machine learning techniques, such as Population Based Training, in the context of a simplified version of the game. The main objective is to assess the cooperation between the hand-authored controller and a human controlled agent, with a particular emphasis on the coordination and minimizing errors during runs of fixed time length. A testing method has been designed in order to more accurately evaluate the performance. Through the implementation of a specialized controller utilizing techniques such as Behavior Trees and Decision Trees, the controller was able to successfully accomplish the task of delivering soups. An analysis was made to examine the performance of the hand-authored controller in comparison to the results obtained in other research papers that use the same simplified version of the game. After the results have been obtained, a clear difference was visible. The scripted AI performed better when paired with himself than one created with population based training that was paired with himself. However, the scripted AI was not better than a coupled planning algorithm that was paired with himself, but the computations were easier and faster for the scripted AI. When paired with a human, the overall performance decreases, but increases only for one specific map., CSE3000 Research Project, Computer Science and Engineering

We propose a new power amplifier (PA) behavioral modeling approach, to characterize and compensate for the signal quality degrading effects induced by a PA with a machine learning (ML) aided piece-wise (PW) modeling approach. Instead of using a single pruned Volterra model, we use multiple small-size pruned Volterra models by classifying the input data into different classes. For that purpose, an ML classifier model is trained by extracting some crucial features from both the input signal statistics and the PA operating point. The simulation results indicate that our approach contributes to an improved performance/complexity trade-off than a single generalized memory polynomial (GMP) model in terms of PA behavior modeling and linearization.

One of the qualities of higher education is determined by the success rate of student learning. Assessment of student success rates is based on student graduation on time. Sultan Syarif Kasim State Islamic University Riau is one of the state universities in Riau, with a total of 30,000 students. Of all the active students, there are some who are not. Students who are not active in this case will affect the timeliness of their graduation. The university always evaluates the performance of its students to find out information related to the factors that cause students to become inactive so that they are more likely to drop out and what data affect students being able to graduate on time. The evaluation results are stored in an academic database so that the data can later be used as supporting data when making decisions by the university. This research used data science concepts to explore and extract data sets from databases to find models or patterns, as well as new insights that can be used as tools for decision-making. After the data was explored, machine learning concepts were used to identify and classify the data. The method used was the Decision Tree Method. The results of the study found that these two concepts can provide the expected results. Based on the test results, it is known that the attribute that influences the success of student studies is the grade point average (GPA), where the accuracy of the maximum recognition rate is 88.19%. Keywords : Data science; Decision Tree; Graduate on Time; Machine Learning

One of the qualities of higher education is determined by the success rate of student learning. Assessment of student success rates is based on student graduation on time. Sultan Syarif Kasim State Islamic University Riau is one of the state universities in Riau, with a total of 30,000 students. Of all the active students, there are some who are not. Students who are not active in this case will affect the timeliness of their graduation. The university always evaluates the performance of its students to find out information related to the factors that cause students to become inactive so that they are more likely to drop out and what data affect students being able to graduate on time. The evaluation results are stored in an academic database so that the data can later be used as supporting data when making decisions by the university. This research used data science concepts to explore and extract data sets from databases to find models or patterns, as well as new insights that can be used as tools for decision-making. After the data was explored, machine learning concepts were used to identify and classify the data. The method used was the Decision Tree Method. The results of the study found that these two concepts can provide the expected results. Based on the test results, it is known that the attribute that influences the success of student studies is the grade point average (GPA), where the accuracy of the maximum recognition rate is 88.19%. Keywords : Data science; Decision Tree; Graduate on Time; Machine Learning

This paper analyzes data from coronary artery bypass grafting (CABG) using decision functions to represent rules. The data was collected at the University Hospital in Umeå, Sweden. The data contains pre-, intra-, and postoperative detail from 2975 heart operations during 1993-96. Each instance is represented by 14 preoperative variables, 4 intraoperative variables, and 9 postoperative variables. A logical method is used to predict the postoperative variables using preoperative variables. First, each postoperative variable is represented as a decision functions of preoperative variables. Then, for each postoperative variable, a minimal set of preoperative variables is derived. And finally, each postoperative variable is represented by a minimum set of rules using preoperative variables. With this method we can predict postoperative outcome, where prediction using preoperative data only is of particular interest e.g. for surgery scheduling.

The present research work focused on predicting the electrical conductivity (EC) of surface water in the Upper Ganga basin using four machine learning algorithms: multilayer perceptron (MLP), co-adaptive neuro-fuzzy inference system (CANFIS), random forest (RF), and decision tree (DT). The study also utilized the gamma test for selecting appropriate input and output combinations. The results of the gamma test revealed that total hardness (TH), magnesium (Mg), and chloride (Cl) parameters were suitable input variables for EC prediction. The performance of the models was evaluated using statistical indices such as Percent Bias (PBIAS), correlation coefficient (R), Willmott’s index of agreement (WI), Index of Agreement (PI), root mean square error (RMSE) and Legate-McCabe Index (LMI). Comparing the results of the EC models using these statistical indices, it was observed that the RF model outperformed the other algorithms. During the training period, the RF algorithm has a small positive bias (PBIAS = 0.11) and achieves a high correlation with the observed values (R = 0.956). Additionally, it shows a low RMSE value (360.42), a relatively good coefficient of efficiency (CE = 0.932), PI (0.083), WI (0.908) and LMI (0.083). However, during the testing period, the algorithm’s performance shows a small negative bias (PBIAS = − 0.46) and a good correlation (R = 0.929). The RMSE value decreases significantly (26.57), indicating better accuracy, the coefficient of efficiency remains high (CE = 0.915), PI (0.033), WI (0.965) and LMI (− 0.028). Similarly, the performance of the RF algorithm during the training and testing periods in Prayagraj. During the training period, the RF algorithm shows a PBIAS of 0.50, indicating a small positive bias. It achieves an RMSE of 368.3, R of 0.909, CE of 0.872, PI of 0.015, WI of 0.921, and LMI of 0.083. During the testing period, the RF algorithm demonstrates a slight negative bias with a PBIAS of − 0.06. The RMSE reduces significantly to 2, Validerad;2023;Nivå 2;2023-09-12 (hanlid);Funder: Deanship of Scientifc Research, King Saud University

Landslides are a dangerous natural hazard that can critically harm road infrastructure in mountainous places, resulting in significant damage and fatalities. The primary purpose of this study was to assess the efficacy of three machine learning algorithms (MLAs) for landslide susceptibility mapping including random forest (RF), decision tree (DT), and support vector machine (SVM). We selected a case study region that is frequently affected by landslides, the important Kamyaran–Sarvabad road in the Kurdistan province of Iran. Altogether, 14 landslide evaluation factors were input into the MLAs including slope, aspect, elevation, river density, distance to river, distance to fault, fault density, distance to road, road density, land use, slope curvature, lithology, stream power index (SPI), and topographic wetness index (TWI). We identified 64 locations of landslides by field survey of which 70% were randomly employed for building and training the three MLAs while the remaining locations were used for validation. The area under the receiver operating characteristics (AUC) reached a value of 0.94 for the decision tree compared to 0.82 for the random forest, and 0.75 for support vector machines model. Thus, the decision tree model was most accurate in identifying the areas at risk for future landslides. The obtained results may inform geoscientists and those in decision-making roles for landslide management., Validerad;2023;Nivå 2;2023-06-19 (hanlid);Funder: University of Kurdistan, Iran (02-9-3786, 01-9-22595)

Permanent magnet synchronous machines have various advantages and have showed the most superiorperformance for Electric Vehicles. However, modeling them is difficult because of their nonlinearity. In orderto deal with the complexity, the artificial neural network and machine learning models including k-nearest neighbors, decision tree, random forest, and multiple linear regression with a quadratic model are developed to predict electrical parameters and losses as new prediction approaches for the performance of Volvo Cars’ electric vehicles and evaluate their performance. The test operation data of the Volvo Cars Corporation was used to extract and calculate the input and output data for each prediction model. In order to smooth the effects of each input variable, the input data was normalized. In addition, correlation matrices of normalized inputs were produced, which showed a high correlation between rotor temperature and winding resistance in the electrical parameter prediction dataset. They also demonstrated a strong correlation between the winding temperature and the rotor temperature in the loss prediction dataset.Grid search with 5-fold cross validation was implemented to optimize hyperparameters of artificial neuralnetwork and machine learning models. The artificial neural network models performed the best in MSE and R-squared scores for all the electrical parameters and loss prediction. The results indicate that artificial neural networks are more successful at handling complicated nonlinear relationships like those seen in electrical systems compared with other machine learning algorithms. Compared to other machine learning algorithms like decision trees, k-nearest neighbors, and multiple linear regression with a quadratic model, random forest produced superior results. With the exception of q-axis voltage, the decision tree model outperformed the knearestneighbors model in terms of parameter prediction, as measured by MSE and R-squared score. Multiple lin

The aim of this study was to find strains in embedded reinforcement by monitoring surface deformations. Compared with analytical methods, application of the machine learning regression technique imparts a noteworthy reduction in modeling complexity caused by the tension stiffening effect. The present research aimed to achieve a hybrid learning approach for non-contact prediction of embedded strains based on surface deformations monitored by digital image correlation (DIC). However, due to the small training dataset collected by the installed strain gauges, the input dataset was enriched by a semi-empirical equation proposed in a previous study. Therefore, the present study discussed (i) instrumentation by strain gauge and DIC as well as data acquisition and post-processing of the data, accounting for strain gradients on the concrete surface and embedded reinforcement; (ii) input dataset generation for training machine learning regression models approaching hybrid learning; (iii) data regression to predict strains in embedded reinforcement based on monitored surface deformations; and (iv) the results, validation, and post-processing responses to make the method more robust. Finally, the developed model was evaluated through numerous statistical performance measures. The results showed that the proposed method can reasonably predict strain in embedded reinforcement, providing an innovative type of sensing application with highly improved performance., Validerad;2023;Nivå 2;2023-04-11 (hanlid)

Permanent magnet synchronous machines have various advantages and have showed the most superiorperformance for Electric Vehicles. However, modeling them is difficult because of their nonlinearity. In orderto deal with the complexity, the artificial neural network and machine learning models including k-nearest neighbors, decision tree, random forest, and multiple linear regression with a quadratic model are developed to predict electrical parameters and losses as new prediction approaches for the performance of Volvo Cars’ electric vehicles and evaluate their performance. The test operation data of the Volvo Cars Corporation was used to extract and calculate the input and output data for each prediction model. In order to smooth the effects of each input variable, the input data was normalized. In addition, correlation matrices of normalized inputs were produced, which showed a high correlation between rotor temperature and winding resistance in the electrical parameter prediction dataset. They also demonstrated a strong correlation between the winding temperature and the rotor temperature in the loss prediction dataset.Grid search with 5-fold cross validation was implemented to optimize hyperparameters of artificial neuralnetwork and machine learning models. The artificial neural network models performed the best in MSE and R-squared scores for all the electrical parameters and loss prediction. The results indicate that artificial neural networks are more successful at handling complicated nonlinear relationships like those seen in electrical systems compared with other machine learning algorithms. Compared to other machine learning algorithms like decision trees, k-nearest neighbors, and multiple linear regression with a quadratic model, random forest produced superior results. With the exception of q-axis voltage, the decision tree model outperformed the knearestneighbors model in terms of parameter prediction, as measured by MSE and R-squared score. Multiple lin

This thesis aims to assist in the development of machine learning models tailored for stress testing. The main objective is to create models that can predict loan defaults while considering the impact of macroeconomic stress. By achieving this, Nordea can continue the development of machine learning models for stress testing by utilizing the models as a basis for further advancement. The research begins with an analysis of historical loan data, encompassing diverse customer and macroeconomic variables that influence loan default rates. Leveraging machine learning algorithms, feature selection methods, data imbalance management and model training techniques, a set of predictive models is constructed. These models aim to capture the intricate relationships between the identified variables and loan defaults, ensuring their suitability for stress testing purposes. The subsequent phase of the research focuses on subjecting the developed models to simulated adverse economic conditions during stress testing. By evaluating the models’ performance under various stressed scenarios, their ability to provide predictions is assessed. This stress testing process allows us to analyse the models’ capabilities of incorporating a stressed scenario in their predictions. The thesis concludes with an evaluation of the developed machine learning models and their abilities to identify defaulted loans in a stressed macroeconomy. By creating these models specifically tailored for stress testing loans, we will provide a basis for further development within the area of stress testing modeling., Denna uppsats syftar till att bidra till utvecklingen av maskininlärningsmodeller lämpade för stress testing. Det främsta målet är att skapa modeller som kan förutsäga lån som kommer att misslyckas samtidigt som de beaktar påverkan av makroekonomisk stress. Genom att uppnå detta kan Nordea fortsätta utvecklingen av maskininlärningsmodeller för stress testning genom att använda modellerna som grund för ytterligare utveckling. Arbetet inleds med en analys av historisk lånedata, som omfattar olika kund- och makroekonomiska variabler som påverkar lån. Genom att använda oss av maskininlärningsalgoritmer, metoder för urval av förklarande variabler, hantering av dataobalans och tekniker för modellträning konstrueras en uppsättning prediktiva modeller. Dessa modeller syftar till att fånga de komplexa relationerna mellan de identifierade variablerna och låneavvikelser och säkerställa deras lämplighet för stress testning. Den efterföljande fasen av arbetet fokuserar på att utsätta de utvecklade modellerna för simulerade stressade ekonomiska förhållanden. Genom att utvärdera modellernas prestanda under olika stressade förhållanden bedöms deras förmåga att prediktera uteblivna lån. Denna process för stress testning gör det möjligt för oss att analysera modellernas förmåga att inkludera stressade förhållanden i sina prediktioner. Uppsatsen avslutas med en utvärdering av de utvecklade maskininlärningsmodellerna och deras förmåga att identifiera uteblivna lån i en stressad makroekonomi. Genom att skapa dessa modeller specifikt anpassade för stresstestning av lån kommer vi att ge en grund för ytterligare utveckling inom området.

Blockchain is a decentralized database that is distributed among a computer network. To enable a smooth decision making process without any authority, different blockchain applications use their own consensus algorithms. The problem is that for a new blockchain application, there is limited aid in deciding which algorithm it should implement. Selecting consensus algorithms is crucial because reaching consensus is the fundamental issue of a decentralized system. Different algorithms are designed with their own advantages and limitations, making it complex to navigate one’s way through a list of consensus algorithms. This thesis attempts to contribute to solving this problem by surveying 15 existing cryptocurrencies’ consensus algorithms used in their blockchain application and then producing a decision tree as the aid for algorithm selection. The top 5 algorithms from each category in Proof of Work (PoW), Proof of Stake (PoS), and Hybrid Proof of Work + Proof of Stake (PoW + PoS) are selected. The research method is qualitative. The study shows that different consensus algorithms often share some properties, but they are usually built to solve the issues of another algorithm, which means they also have their own distinctive advantages. Therefore, the decision tree reveals how these algorithms are logically connected and the key properties blockchain consensus algorithms possess. Based on the result of this thesis, further research can be conducted to include more algorithms in order to make the decision tree more comprehensive. Implementations of these algorithms in similar network setup can also be done to experiment with their claimed properties. The decision tree can be sent to industry for further feedback., Blockchain är en decentraliserad databas som distribueras i ett datornätverk. För att möjliggöra en smidig beslutsprocess utan någon auktoritet använder olika blockkedjeapplikationer sina egna konsensusalgoritmer. Problemet är att för en ny blockchain-applikation finns det begränsad hjälp för att bestämma vilken algoritm den ska implementera. Att välja konsensusalgoritmer är avgörande eftersom att nå konsensus är den grundläggande frågan för ett decentraliserat system. Olika algoritmer är designade med sina egna fördelar och begränsningar, vilket gör det komplicerat att navigera sig igenom en lista med konsensusalgoritmer. Forskningsmetoden är kvalitativ. Det här dokumentet försöker bidra till att lösa detta problem genom att kartlägga 15 befintliga kryptovalutors konsensusalgoritmer som används i deras blockkedjeapplikation och sedan ta fram ett beslutsträd som hjälp för val av algoritmer. De 5 bästa algoritmerna från varje kategori i Proof of Work (PoW), Proof of Stake (PoS) och Hybrid Proof of Work + Proof of Stake (PoW + PoS) väljs. Studien visar att olika konsensusalgoritmer ofta delar vissa egenskaper, men de är vanligtvis byggda för att lösa problem med en annan algoritm, vilket innebär att de också har sina egna distinkta fördelar. Därför avslöjar beslutsträdet hur dessa algoritmer är logiskt kopplade och de nyckelegenskaper som blockchain konsensusalgoritmer besitter. Baserat på resultatet av denna artikel kan ytterligare forskning utföras för att inkludera fler algoritmer för att göra beslutsträdet mer heltäckande. Implementeringar av dessa algoritmer i liknande nätverksuppsättningar kan också göras för att experimentera med deras påstådda egenskaper. Beslutsträdet kan skickas till industrin för vidare feedback.

In recent years the development and implementation of AI and machine learning models has increased dramatically. The availability of quality data paving the way for sophisticated AI models. Financial institutions uses many models in their daily operations. They are however, heavily regulated and need to follow the regulation that are set by central banks auditory standard and the financial supervisory authorities. One of these standards is the disclosure of expected credit losses in financial statements of banks, called IFRS 9. Banks must measure the expected credit shortfall in line with regulations set up by the EBA and FSA. In this master thesis, we are collaborating with a Swedish bank to evaluate different machine learning models to predict defaults of a unsecured credit portfolio. The default probability is a key variable in the expected credit loss equation. The goal is not only to develop a valid model to predict these defaults but to create and evaluate different models based on their performance and transparency. With regulatory challenges within AI the need to introduce transparency in models are part of the process. When banks use models there’s a requirement on transparency which refers to of how easily a model can be understood with its architecture, calculations, feature importance and logic’s behind the decision making process. We have compared the commonly used model logistic regression to three machine learning models, decision tree, random forest and XG boost. Where we want to show the performance and transparency differences of the machine learning models and the industry standard. We have introduced a transparency evaluation tool called transparency matrix to shed light on the different transparency requirements of machine learning models. The results show that all of the tree based machine learning models are a better choice of algorithm when estimating defaults compared to the traditional logistic regression. This is shown in the AUC score as, Under de senaste ̊aren har utvecklingen och implementeringen av AI- och maskininl ̈arningsmodeller o ̈kat dramatiskt. Tillg ̊angen till kvalitetsdata banar va ̈gen fo ̈r sofistikerade AI-modeller. Finansiella institutioner anva ̈nder m ̊anga modeller i sin dagliga verksamhet. De a ̈r dock starkt reglerade och m ̊aste fo ̈lja de regler som faststa ̈lls av centralbankernas revisionsstandard och finansiella tillsynsmyndigheter. En av dessa standarder a ̈r offentligg ̈orandet av fo ̈rva ̈ntade kreditfo ̈rluster i bankernas finansiella rapporter, kallad IFRS 9. Banker m ̊aste ma ̈ta den fo ̈rva ̈ntade kreditfo ̈rlusten i linje med regler som faststa ̈lls av EBA och FSA. I denna uppsats samarbetar vi med en svensk bank fo ̈r att utva ̈rdera olika maskininl ̈arningsmodeller f ̈or att fo ̈rutsa ̈ga fallisemang i en blankokreditsportfo ̈lj. Sannolikheten fo ̈r fallismang ̈ar en viktig variabel i ekvationen fo ̈r fo ̈rva ̈ntade kreditfo ̈rluster. M ̊alet a ̈r inte bara att utveckla en bra modell fo ̈r att prediktera fallismang, utan ocks ̊a att skapa och utva ̈rdera olika modeller baserat p ̊a deras prestanda och transparens. Med de utmaningar som finns inom AI a ̈r behovet av att info ̈ra transparens i modeller en del av processen. Na ̈r banker anva ̈nder modeller finns det krav p ̊a transparens som ha ̈nvisar till hur enkelt en modell kan fo ̈rst ̊as med sin arkitektur, bera ̈kningar, variabel p ̊averkan och logik bakom beslutsprocessen. Vi har ja ̈mfo ̈rt den vanligt anva ̈nda modellen logistisk regression med tre maskininla ̈rningsmodeller: Decision trees, Random forest och XG Boost. Vi vill visa skillnaderna i prestanda och transparens mellan maskininl ̈arningsmodeller och branschstandarden. Vi har introducerat ett verktyg fo ̈r transparensutva ̈rdering som kallas transparensmatris fo ̈r att belysa de olika transparenskraven fo ̈r maskininla ̈rningsmodeller. Resultaten visar att alla tra ̈d-baserade maskininla ̈rningsmodeller a ̈r ett ba ̈ttre val av modell vid prediktio

Today it is important for investors to identify which stocks that will result in positive returns in order for the right decision to be made when trading on the stock market. For decades it has been an area of interest for academics, and it is still challenging due to many difficulties and problems. A large number of studies has been carried out in machine learning and stock trading,where many of the studies has resulted in promising results despite these challenges. The aim of this study was to develop and evaluate predictive models for identifying stocks that outperform the Swedish market index OMXSPI. The research utilized a dataset of historical stock data and applied three various machine learning algorithms, Support Vector Machine, Logistic Regression and Decision Trees to predict if excess performance was met. With the help of ten-fold cross-validation and hyperparameter tuning the results were an IT-artefact that produced satisfying results. The results showed that hyperparameter tuning techniques marginally improved the metrics focused-on, namely accuracy and precision. The support vector machine model achieved an accuracy of 58,52% and a precision of 57,51%. The logistic regression model achieved an accuracy of 55,75% and a precision of 54,81%. Finally, the decision tree model which was the best performer, achieved an accuracy of 64,84% and a precision of 65,00%.

This thesis is a cooperation with Knowit, Östrand \& Hansen, and Orkla. It aimed to explore the application of Machine Learning and Deep Learning models with Conformal Prediction for a predictive maintenance situation at Orkla. Predictive maintenance is essential in numerous industrial manufacturing scenarios. It can help to reduce machine downtime, improve equipment reliability, and save unnecessary costs. In this thesis, various Machine Learning and Deep Learning models, including Decision Tree, Random Forest, Support Vector Regression, Gradient Boosting, and Long short-term memory, are applied to a real-world predictive maintenance dataset. The Orkla dataset was originally planned to use in this thesis project. However, due to some challenges met and time limitations, one NASA C-MAPSS dataset with a similar data structure was chosen to study how Machine Learning models could be applied to predict the remaining useful lifetime (RUL) in manufacturing. Besides, conformal prediction, a recently developed framework to measure the prediction uncertainty of Machine Learning models, is also integrated into the models for more reliable RUL prediction. The thesis project results show that both the Machine Learning and Deep Learning models with conformal prediction could predict RUL closer to the true RUL while LSTM outperforms the Machine Learning models. Also, the conformal prediction intervals provide informative and reliable information about the uncertainty of the predictions, which can help inform personnel at factories in advance to take necessary maintenance actions. Overall, this thesis demonstrates the effectiveness of utilizing machine learning and Deep Learning models with Conformal Prediction for predictive maintenance situations. Moreover, based on the modeling results of the NASA dataset, some insights are discussed on how to transfer these experiences into Orkla data for RUL prediction in the future.

This thesis investigates the pair production of Higgs bosons (di-Higgs events) at the ATLAS experiment in the Large Hadron Collider (LHC), focusing on the channel where one Higgs boson decays into two bottom quarks and the other decays into two tau leptons. The main objective was to determine whether introducing a split in the invariant mass of the decay products from the two Higgs bosons (the di-Higgs mass) and using this as an analysis variable improves the sensitivity of the Boosted Decision Tree (BDT) machine learning algorithm to the di-Higgs signal. A mass split was performed at 350 GeV, and the BDT algorithm was trained on both the split and un-split data sets, where the split data set included a high-mass region (di-Higgs mass above 350 GeV) using the Standard Model Higgs boson coupling constant of 1 and a low-mass region (di-Higgs mass below 350 GeV) using the enhanced coupling constant of 10 to create a low-mass region more sensitive to the signal. The results showed that the BDT algorithm training performed on the split data set provided a 3.6% improvement in the exclusion limits, indicating an improvement in the algorithm's sensitivity to the di-Higgs signal compared to the training performed on the un-split data set. This finding suggests that the introduction of a split at 350 GeV can enhance the accuracy and efficiency of machine learning algorithms in detecting di-Higgs boson production at the LHC. The improvement in sensitivity was attributed to the enhanced discrimination between signal and background events provided by the split in the di-Higgs mass analysis variable. The improved separation between the signal and background events lead to a higher signal-to-background ratio and a corresponding increase in the BDT algorithm's sensitivity to the di-Higgs signal. In conclusion, this thesis provided evidence that introducing a split in the di-Higgs mass analysis variable can improve the sensitivity of machine learning algorithms to the di-Higgs si

Blockchain is a decentralized database that is distributed among a computer network. To enable a smooth decision making process without any authority, different blockchain applications use their own consensus algorithms. The problem is that for a new blockchain application, there is limited aid in deciding which algorithm it should implement. Selecting consensus algorithms is crucial because reaching consensus is the fundamental issue of a decentralized system. Different algorithms are designed with their own advantages and limitations, making it complex to navigate one’s way through a list of consensus algorithms. This thesis attempts to contribute to solving this problem by surveying 15 existing cryptocurrencies’ consensus algorithms used in their blockchain application and then producing a decision tree as the aid for algorithm selection. The top 5 algorithms from each category in Proof of Work (PoW), Proof of Stake (PoS), and Hybrid Proof of Work + Proof of Stake (PoW + PoS) are selected. The research method is qualitative. The study shows that different consensus algorithms often share some properties, but they are usually built to solve the issues of another algorithm, which means they also have their own distinctive advantages. Therefore, the decision tree reveals how these algorithms are logically connected and the key properties blockchain consensus algorithms possess. Based on the result of this thesis, further research can be conducted to include more algorithms in order to make the decision tree more comprehensive. Implementations of these algorithms in similar network setup can also be done to experiment with their claimed properties. The decision tree can be sent to industry for further feedback., Blockchain är en decentraliserad databas som distribueras i ett datornätverk. För att möjliggöra en smidig beslutsprocess utan någon auktoritet använder olika blockkedjeapplikationer sina egna konsensusalgoritmer. Problemet är att för en ny blockchain-applikation finns det begränsad hjälp för att bestämma vilken algoritm den ska implementera. Att välja konsensusalgoritmer är avgörande eftersom att nå konsensus är den grundläggande frågan för ett decentraliserat system. Olika algoritmer är designade med sina egna fördelar och begränsningar, vilket gör det komplicerat att navigera sig igenom en lista med konsensusalgoritmer. Forskningsmetoden är kvalitativ. Det här dokumentet försöker bidra till att lösa detta problem genom att kartlägga 15 befintliga kryptovalutors konsensusalgoritmer som används i deras blockkedjeapplikation och sedan ta fram ett beslutsträd som hjälp för val av algoritmer. De 5 bästa algoritmerna från varje kategori i Proof of Work (PoW), Proof of Stake (PoS) och Hybrid Proof of Work + Proof of Stake (PoW + PoS) väljs. Studien visar att olika konsensusalgoritmer ofta delar vissa egenskaper, men de är vanligtvis byggda för att lösa problem med en annan algoritm, vilket innebär att de också har sina egna distinkta fördelar. Därför avslöjar beslutsträdet hur dessa algoritmer är logiskt kopplade och de nyckelegenskaper som blockchain konsensusalgoritmer besitter. Baserat på resultatet av denna artikel kan ytterligare forskning utföras för att inkludera fler algoritmer för att göra beslutsträdet mer heltäckande. Implementeringar av dessa algoritmer i liknande nätverksuppsättningar kan också göras för att experimentera med deras påstådda egenskaper. Beslutsträdet kan skickas till industrin för vidare feedback.

The present research work focused on predicting the electrical conductivity (EC) of surface water in the Upper Ganga basin using four machine learning algorithms: multilayer perceptron (MLP), co-adaptive neuro-fuzzy inference system (CANFIS), random forest (RF), and decision tree (DT). The study also utilized the gamma test for selecting appropriate input and output combinations. The results of the gamma test revealed that total hardness (TH), magnesium (Mg), and chloride (Cl) parameters were suitable input variables for EC prediction. The performance of the models was evaluated using statistical indices such as Percent Bias (PBIAS), correlation coefficient (R), Willmott’s index of agreement (WI), Index of Agreement (PI), root mean square error (RMSE) and Legate-McCabe Index (LMI). Comparing the results of the EC models using these statistical indices, it was observed that the RF model outperformed the other algorithms. During the training period, the RF algorithm has a small positive bias (PBIAS = 0.11) and achieves a high correlation with the observed values (R = 0.956). Additionally, it shows a low RMSE value (360.42), a relatively good coefficient of efficiency (CE = 0.932), PI (0.083), WI (0.908) and LMI (0.083). However, during the testing period, the algorithm’s performance shows a small negative bias (PBIAS = − 0.46) and a good correlation (R = 0.929). The RMSE value decreases significantly (26.57), indicating better accuracy, the coefficient of efficiency remains high (CE = 0.915), PI (0.033), WI (0.965) and LMI (− 0.028). Similarly, the performance of the RF algorithm during the training and testing periods in Prayagraj. During the training period, the RF algorithm shows a PBIAS of 0.50, indicating a small positive bias. It achieves an RMSE of 368.3, R of 0.909, CE of 0.872, PI of 0.015, WI of 0.921, and LMI of 0.083. During the testing period, the RF algorithm demonstrates a slight negative bias with a PBIAS of − 0.06. The RMSE reduces significantly to 2, Validerad;2023;Nivå 2;2023-09-12 (hanlid);Funder: Deanship of Scientifc Research, King Saud University

Landslides are a dangerous natural hazard that can critically harm road infrastructure in mountainous places, resulting in significant damage and fatalities. The primary purpose of this study was to assess the efficacy of three machine learning algorithms (MLAs) for landslide susceptibility mapping including random forest (RF), decision tree (DT), and support vector machine (SVM). We selected a case study region that is frequently affected by landslides, the important Kamyaran–Sarvabad road in the Kurdistan province of Iran. Altogether, 14 landslide evaluation factors were input into the MLAs including slope, aspect, elevation, river density, distance to river, distance to fault, fault density, distance to road, road density, land use, slope curvature, lithology, stream power index (SPI), and topographic wetness index (TWI). We identified 64 locations of landslides by field survey of which 70% were randomly employed for building and training the three MLAs while the remaining locations were used for validation. The area under the receiver operating characteristics (AUC) reached a value of 0.94 for the decision tree compared to 0.82 for the random forest, and 0.75 for support vector machines model. Thus, the decision tree model was most accurate in identifying the areas at risk for future landslides. The obtained results may inform geoscientists and those in decision-making roles for landslide management., Validerad;2023;Nivå 2;2023-06-19 (hanlid);Funder: University of Kurdistan, Iran (02-9-3786, 01-9-22595)

The aim of this study was to find strains in embedded reinforcement by monitoring surface deformations. Compared with analytical methods, application of the machine learning regression technique imparts a noteworthy reduction in modeling complexity caused by the tension stiffening effect. The present research aimed to achieve a hybrid learning approach for non-contact prediction of embedded strains based on surface deformations monitored by digital image correlation (DIC). However, due to the small training dataset collected by the installed strain gauges, the input dataset was enriched by a semi-empirical equation proposed in a previous study. Therefore, the present study discussed (i) instrumentation by strain gauge and DIC as well as data acquisition and post-processing of the data, accounting for strain gradients on the concrete surface and embedded reinforcement; (ii) input dataset generation for training machine learning regression models approaching hybrid learning; (iii) data regression to predict strains in embedded reinforcement based on monitored surface deformations; and (iv) the results, validation, and post-processing responses to make the method more robust. Finally, the developed model was evaluated through numerous statistical performance measures. The results showed that the proposed method can reasonably predict strain in embedded reinforcement, providing an innovative type of sensing application with highly improved performance., Validerad;2023;Nivå 2;2023-04-11 (hanlid)

Scoliosis is an abnormal curvature of the spine and often diagnosed in childhood or early adolescence. In this study, the risk factors for scoliosis in elementary school children is investigate based on age, backpack weight and gender. There are 260 children participated in this study from aged 7 up to 12 years old. Scoliometer is used to measure the angle of trunk rotation (ATR) on Adam Forward Bending Test. Statistical analysis of analysis of variance (ANOVA) is used to determine the characteristic difference of ATR readings on the risk factors for scoliosis. Significant results with P-value less than 0.001 are found among ATR readings on a linear combination of risk factors for scoliosis of age and backpack weight. Then, the risk factors for scoliosis are classified among elementary school children using Decision Tree and K-Nearest Neighbor. The classification results shown that both Decision Tree method produced highest classification percentage up to 98.08%. This finding indicates that age and backpack weight are significant as the risk factors for scoliosis.

Hate speech is an issue that frequently happens when somebody speaks with one another utilizing social media on the Web. Despite the fact that artificial intelligence frameworks are set up to ban such messages. The main issue that arises in these cases is the high false positive rates, so we need mechanisms to help counter such issues and distinguish hate speech without sabotaging the freedom of expression. Likewise, the identification of hate speech in recordings is more difficult than straightforward plain text. In this project we'll research and make a system for detection of hate speech in videos and text utilizing a classification approach in AI with different techniques such as Logistic Regression, SVM, decision tree. The execution of video to text is also something we'll work on. The recent uptick of people using social media has led to a lot of conflicting views, although healthy conversation is an important part of debates some of the content that is being shared online is offensive and can often be downright hateful. It is important to recognize which messages are actually hateful and offensive and which are not to encourage healthy debates over the internet.

A very challenging and difficult part of finding a cure for Rheumatoid Arthritis ( RA) disease is the same as finding a cure for an autoimmune disease. In the case of autoimmune diseases, it is very difficult to find the cause. RA is also one of the autoimmune diseases that causes inflammation in joints. It does not only limit to joints but also can spread or occur in various parts of the body like skin, lungs, and many others. Many features like age, sex, and previous history of certain diseases affect the occurrence of RA. The proposed ML model uses a decision tree algorithm to analyze a comprehensive set of clinical and laboratory variables collected from RA patients. These variables include age, ID, time, treatment, gender, and other baseline features. The ML model employs feature selection techniques to identify the most relevant predictors that contribute to disease progression. By training the decision tree algorithm on a large dataset of RA patients, the model generates an accurate predictive model for assessing disease severity and progression. The model is able to detect RA disease with almost 80% accuracy with the given training dataset. RA can have many other symptoms and characteristics. This data was trained on a basic dataset available from the Kaggle website. The dataset is in CSV format and has 6 features. The model was trained using a decision tree algorithm, which categorizes the data into different categories and then checks for specific data when needed in Jupyter Notebook. The study explored detecting basic symptoms and characteristics of RA disease. The Decision tree model has given accuracy of 86.5% and 74.4% on training and testing respectively. So Accuracy of the model is pretty good for the dataset used.

Due to significant technology advancements, people's needs have expanded. As a result, there have been more requests for loan approval in the banking sector. A few qualities, taken for consideration, when choosing a candidate for loan approval in order to, determine loan's status. Banks face a major challenge; when it, comes to assessing loan applications and lowering the risks associated with potential borrower defaults. Since they must thoroughly evaluate each borrower's eligibility for a loan, banks find this process to be particularly challenging. This research proposes combining machine learning (ML) models and ensemble learning approaches to find the probability of accepting individual loan requests. This tactic can increase the accuracy with which qualified candidates are selected from a pool of applicants. As a result, this method can be used to address the problems with loan approval processes outlined above. Both the loan applicants and the bank employees profit from the strategy's dramatic reduction in sanctioning time. Because of the banking industry's expansion, more people were applying to loans at banks. In order to predict the accuracy of loan approval status for applied person, we used four different algorithms namely Random Forest, Naive Bayes, Decision Tree, and KNN. By using these, we obtained better accuracy of 83.73% with Naïve Bayes algorithm as best one.

Cancer is a disease characterized by abnormal cell growth and is not contagious, such as breast cancer which can affect both men and women. breast cancer is one of the cancer diseases that is classified as dangerous and takes many victims. However, the biggest problem in this study is that the classification method is low and the resulting accuracy is less than optimal. the purpose of this study is to improve the accuracy of breast cancer classification. Therefore, a new method is proposed, namely ensemble learning which combines logistic regression, decision tree, and random forest methods, with a voting system. This system is useful for finding the best results on each parameter that will produce the best prediction accuracy. The prediction results from this method reached an accuracy of 98.24%. The resulting accuracy rate is more optimal by using the proposed method.

Machine Learning is used across many ranges around the world. The healthcare industry is no exclusion. Machine Learning can play an essential role in predicting presence/absence of locomotors disorders, heart diseases and more. Such information, if predicted well in advance, can provide important intuitions to doctors who can then adapt their diagnosis and dealing per patient basis. We work on predicting possible heart diseases in people using Machine Learning algorithms. In this project we perform the comparative analysis of classifiers like decision tree, Naïve Bayes, Logistic Regression, SVM and Random Forest and we propose an ensemble classifier which perform hybrid classification by taking strong and weak classifiers since it can have multiple number of samples for training and validating the data so we perform the analysis of existing classifier and proposed classifier like Ada-boost and XG-boost which can give the better accuracy and predictive analysis.

Today it is important for investors to identify which stocks that will result in positive returns in order for the right decision to be made when trading on the stock market. For decades it has been an area of interest for academics, and it is still challenging due to many difficulties and problems. A large number of studies has been carried out in machine learning and stock trading,where many of the studies has resulted in promising results despite these challenges. The aim of this study was to develop and evaluate predictive models for identifying stocks that outperform the Swedish market index OMXSPI. The research utilized a dataset of historical stock data and applied three various machine learning algorithms, Support Vector Machine, Logistic Regression and Decision Trees to predict if excess performance was met. With the help of ten-fold cross-validation and hyperparameter tuning the results were an IT-artefact that produced satisfying results. The results showed that hyperparameter tuning techniques marginally improved the metrics focused-on, namely accuracy and precision. The support vector machine model achieved an accuracy of 58,52% and a precision of 57,51%. The logistic regression model achieved an accuracy of 55,75% and a precision of 54,81%. Finally, the decision tree model which was the best performer, achieved an accuracy of 64,84% and a precision of 65,00%.

This thesis aims to assist in the development of machine learning models tailored for stress testing. The main objective is to create models that can predict loan defaults while considering the impact of macroeconomic stress. By achieving this, Nordea can continue the development of machine learning models for stress testing by utilizing the models as a basis for further advancement. The research begins with an analysis of historical loan data, encompassing diverse customer and macroeconomic variables that influence loan default rates. Leveraging machine learning algorithms, feature selection methods, data imbalance management and model training techniques, a set of predictive models is constructed. These models aim to capture the intricate relationships between the identified variables and loan defaults, ensuring their suitability for stress testing purposes. The subsequent phase of the research focuses on subjecting the developed models to simulated adverse economic conditions during stress testing. By evaluating the models’ performance under various stressed scenarios, their ability to provide predictions is assessed. This stress testing process allows us to analyse the models’ capabilities of incorporating a stressed scenario in their predictions. The thesis concludes with an evaluation of the developed machine learning models and their abilities to identify defaulted loans in a stressed macroeconomy. By creating these models specifically tailored for stress testing loans, we will provide a basis for further development within the area of stress testing modeling., Denna uppsats syftar till att bidra till utvecklingen av maskininlärningsmodeller lämpade för stress testing. Det främsta målet är att skapa modeller som kan förutsäga lån som kommer att misslyckas samtidigt som de beaktar påverkan av makroekonomisk stress. Genom att uppnå detta kan Nordea fortsätta utvecklingen av maskininlärningsmodeller för stress testning genom att använda modellerna som grund för ytterligare utveckling. Arbetet inleds med en analys av historisk lånedata, som omfattar olika kund- och makroekonomiska variabler som påverkar lån. Genom att använda oss av maskininlärningsalgoritmer, metoder för urval av förklarande variabler, hantering av dataobalans och tekniker för modellträning konstrueras en uppsättning prediktiva modeller. Dessa modeller syftar till att fånga de komplexa relationerna mellan de identifierade variablerna och låneavvikelser och säkerställa deras lämplighet för stress testning. Den efterföljande fasen av arbetet fokuserar på att utsätta de utvecklade modellerna för simulerade stressade ekonomiska förhållanden. Genom att utvärdera modellernas prestanda under olika stressade förhållanden bedöms deras förmåga att prediktera uteblivna lån. Denna process för stress testning gör det möjligt för oss att analysera modellernas förmåga att inkludera stressade förhållanden i sina prediktioner. Uppsatsen avslutas med en utvärdering av de utvecklade maskininlärningsmodellerna och deras förmåga att identifiera uteblivna lån i en stressad makroekonomi. Genom att skapa dessa modeller specifikt anpassade för stresstestning av lån kommer vi att ge en grund för ytterligare utveckling inom området.

This thesis is a cooperation with Knowit, Östrand \& Hansen, and Orkla. It aimed to explore the application of Machine Learning and Deep Learning models with Conformal Prediction for a predictive maintenance situation at Orkla. Predictive maintenance is essential in numerous industrial manufacturing scenarios. It can help to reduce machine downtime, improve equipment reliability, and save unnecessary costs. In this thesis, various Machine Learning and Deep Learning models, including Decision Tree, Random Forest, Support Vector Regression, Gradient Boosting, and Long short-term memory, are applied to a real-world predictive maintenance dataset. The Orkla dataset was originally planned to use in this thesis project. However, due to some challenges met and time limitations, one NASA C-MAPSS dataset with a similar data structure was chosen to study how Machine Learning models could be applied to predict the remaining useful lifetime (RUL) in manufacturing. Besides, conformal prediction, a recently developed framework to measure the prediction uncertainty of Machine Learning models, is also integrated into the models for more reliable RUL prediction. The thesis project results show that both the Machine Learning and Deep Learning models with conformal prediction could predict RUL closer to the true RUL while LSTM outperforms the Machine Learning models. Also, the conformal prediction intervals provide informative and reliable information about the uncertainty of the predictions, which can help inform personnel at factories in advance to take necessary maintenance actions. Overall, this thesis demonstrates the effectiveness of utilizing machine learning and Deep Learning models with Conformal Prediction for predictive maintenance situations. Moreover, based on the modeling results of the NASA dataset, some insights are discussed on how to transfer these experiences into Orkla data for RUL prediction in the future.

In recent years the development and implementation of AI and machine learning models has increased dramatically. The availability of quality data paving the way for sophisticated AI models. Financial institutions uses many models in their daily operations. They are however, heavily regulated and need to follow the regulation that are set by central banks auditory standard and the financial supervisory authorities. One of these standards is the disclosure of expected credit losses in financial statements of banks, called IFRS 9. Banks must measure the expected credit shortfall in line with regulations set up by the EBA and FSA. In this master thesis, we are collaborating with a Swedish bank to evaluate different machine learning models to predict defaults of a unsecured credit portfolio. The default probability is a key variable in the expected credit loss equation. The goal is not only to develop a valid model to predict these defaults but to create and evaluate different models based on their performance and transparency. With regulatory challenges within AI the need to introduce transparency in models are part of the process. When banks use models there’s a requirement on transparency which refers to of how easily a model can be understood with its architecture, calculations, feature importance and logic’s behind the decision making process. We have compared the commonly used model logistic regression to three machine learning models, decision tree, random forest and XG boost. Where we want to show the performance and transparency differences of the machine learning models and the industry standard. We have introduced a transparency evaluation tool called transparency matrix to shed light on the different transparency requirements of machine learning models. The results show that all of the tree based machine learning models are a better choice of algorithm when estimating defaults compared to the traditional logistic regression. This is shown in the AUC score as, Under de senaste ̊aren har utvecklingen och implementeringen av AI- och maskininl ̈arningsmodeller o ̈kat dramatiskt. Tillg ̊angen till kvalitetsdata banar va ̈gen fo ̈r sofistikerade AI-modeller. Finansiella institutioner anva ̈nder m ̊anga modeller i sin dagliga verksamhet. De a ̈r dock starkt reglerade och m ̊aste fo ̈lja de regler som faststa ̈lls av centralbankernas revisionsstandard och finansiella tillsynsmyndigheter. En av dessa standarder a ̈r offentligg ̈orandet av fo ̈rva ̈ntade kreditfo ̈rluster i bankernas finansiella rapporter, kallad IFRS 9. Banker m ̊aste ma ̈ta den fo ̈rva ̈ntade kreditfo ̈rlusten i linje med regler som faststa ̈lls av EBA och FSA. I denna uppsats samarbetar vi med en svensk bank fo ̈r att utva ̈rdera olika maskininl ̈arningsmodeller f ̈or att fo ̈rutsa ̈ga fallisemang i en blankokreditsportfo ̈lj. Sannolikheten fo ̈r fallismang ̈ar en viktig variabel i ekvationen fo ̈r fo ̈rva ̈ntade kreditfo ̈rluster. M ̊alet a ̈r inte bara att utveckla en bra modell fo ̈r att prediktera fallismang, utan ocks ̊a att skapa och utva ̈rdera olika modeller baserat p ̊a deras prestanda och transparens. Med de utmaningar som finns inom AI a ̈r behovet av att info ̈ra transparens i modeller en del av processen. Na ̈r banker anva ̈nder modeller finns det krav p ̊a transparens som ha ̈nvisar till hur enkelt en modell kan fo ̈rst ̊as med sin arkitektur, bera ̈kningar, variabel p ̊averkan och logik bakom beslutsprocessen. Vi har ja ̈mfo ̈rt den vanligt anva ̈nda modellen logistisk regression med tre maskininla ̈rningsmodeller: Decision trees, Random forest och XG Boost. Vi vill visa skillnaderna i prestanda och transparens mellan maskininl ̈arningsmodeller och branschstandarden. Vi har introducerat ett verktyg fo ̈r transparensutva ̈rdering som kallas transparensmatris fo ̈r att belysa de olika transparenskraven fo ̈r maskininla ̈rningsmodeller. Resultaten visar att alla tra ̈d-baserade maskininla ̈rningsmodeller a ̈r ett ba ̈ttre val av modell vid prediktio

This thesis investigates the pair production of Higgs bosons (di-Higgs events) at the ATLAS experiment in the Large Hadron Collider (LHC), focusing on the channel where one Higgs boson decays into two bottom quarks and the other decays into two tau leptons. The main objective was to determine whether introducing a split in the invariant mass of the decay products from the two Higgs bosons (the di-Higgs mass) and using this as an analysis variable improves the sensitivity of the Boosted Decision Tree (BDT) machine learning algorithm to the di-Higgs signal. A mass split was performed at 350 GeV, and the BDT algorithm was trained on both the split and un-split data sets, where the split data set included a high-mass region (di-Higgs mass above 350 GeV) using the Standard Model Higgs boson coupling constant of 1 and a low-mass region (di-Higgs mass below 350 GeV) using the enhanced coupling constant of 10 to create a low-mass region more sensitive to the signal. The results showed that the BDT algorithm training performed on the split data set provided a 3.6% improvement in the exclusion limits, indicating an improvement in the algorithm's sensitivity to the di-Higgs signal compared to the training performed on the un-split data set. This finding suggests that the introduction of a split at 350 GeV can enhance the accuracy and efficiency of machine learning algorithms in detecting di-Higgs boson production at the LHC. The improvement in sensitivity was attributed to the enhanced discrimination between signal and background events provided by the split in the di-Higgs mass analysis variable. The improved separation between the signal and background events lead to a higher signal-to-background ratio and a corresponding increase in the BDT algorithm's sensitivity to the di-Higgs signal. In conclusion, this thesis provided evidence that introducing a split in the di-Higgs mass analysis variable can improve the sensitivity of machine learning algorithms to the di-Higgs si