Search

The automotive industry continues to grow rapidly with technological developments. Vehicles are being equipped with new software and sensors which assist drivers. People have become familiar with the L1 and L2 vehicles with advanced driver assistance systems. However, the idea of full automation is still new to most people. In Türkiye, AVs are not present on roads. Thus, neither pedestrians nor drivers have been in interaction with AVs yet. This study investigates the preliminary perception of AVs among road users in Türkiye. The data were collected via a face-to-face survey. A binary logit and a random forest model were used to identify the critical factors in the positive attitudes toward AVs. Binary logit and random forest models resulted in similar findings. Age, education level, people who would feel safe driving in an AV, and who believed that urban traffic congestion would decrease were positive about AVs. The random forest model showed the same accuracy when the two models were compared. Most public opinion studies about AVs have been investigated in developed countries. It is believed that this study will help to understand the perception of AVs in developing countries. [ABSTRACT FROM AUTHOR]

As valued members of forest ecosystems, fast-growing tree species are of great importance for protecting natural forests, fulfilling the demand for raw wood material, and preserving biodiversity and ecosystem services. In this regard, multi-temporal monitoring and evaluation of poplar planted areas are essential for decision makers and planners to keep the inventory maps of planted areas up-to-date and track the temporal changes accurately. Using a spectral library, available spectral knowledge can be incorporated into inventory mapping and image classification processes. In this study, in situ spectral measurements of five Populus (four clones and one species) planted sites in Türkiye were collected at three phenological stages (early, mid, and late vegetation) and a novel spectral library was created after completing strict processing stages. Analysis of the spectral reflectances of the Populus species revealed that their spectral properties in the visible region are similar, whereas the discrimination in the near-infrared and shortwave infrared (SWIR) regions was noticeably higher. Two types of applications were conducted to effectively use the spectral library. First, the spectral library was used to validate the poplar class labels of the sites visited during the field campaign. The effectiveness of Worldview-3 imagery with eight SWIR bands was evaluated by pairwise comparison of the spectral curves of the sample fields with those of the spectral library. Second, the spectral library facilitated the extraction of poplar samples from Sentinel-2 imagery in an unvisited region, and a supervised classification process was conducted using the random forest algorithm. The results revealed that the Populus species (Samsun and I-214) could be delineated with high accuracy (>70%) using 10-band Sentinel-2 imagery, and object-based classification outperformed pixel-based classification by approximately 5% for the poplar types. The developed spectral library stands as a valuable asset for studies focused on identifying and classifying Populus species using high-resolution imagery. [ABSTRACT FROM AUTHOR]

Olives are a crucial economic crop in Mediterranean countries. Detailed spatial information on the distribution and condition of crops at regional and national scales is essential to ensure the continuity of crop quality and yield efficiency. However, most earlier studies on olive tree mapping focused mainly on small parcels using single-sensor, very high resolution (VHR) data, which is time-consuming, expensive and cannot feasibly be scaled up to a larger area. Therefore, we evaluated the performance of Sentinel-1 and Sentinel-2 data fusion for the regional mapping of olive trees for the first time, using the Izmir Province of Türkiye, an ancient olive-growing region, as a case study. Three different monthly composite images reflecting the different phenological stages of olive trees were selected to separate olive trees from other land cover types. Seven land-cover classes, including olives, were mapped separately using a random forest classifier for each year between 2017 and 2021. The results were assessed using the k-fold cross-validation method, and the final olive tree map of Izmir was produced by combining the olive tree distribution over two consecutive years. District-level areas covered by olive trees were calculated and validated using official statistics from the Turkish Statistical Institute (TUIK). The K-fold cross-validation accuracy varied from 94% to 95% between 2017 and 2021, and the final olive map achieved 98% overall accuracy with 93% producer accuracy for the olive class. The district-level olive area was strongly related to the TUIK statistics (R2 = 0.60, NRMSE = 0.64). This study used Sentinel data and Google Earth Engine (GEE) to produce a regional-scale olive distribution map that can be scaled up to the entire country and replicated elsewhere. This map can, therefore, be used as a foundation for other scientific studies on olive trees, particularly for the development of effective management practices. [ABSTRACT FROM AUTHOR]

Machine learning approaches are increasingly being employed to forecast the key characteristics of strong ground motions, including the challenging classification of Pulse-Like (PL) ground motions. The PL ground motions are characterized by their impulsive nature and have the potential to cause significant damage to structures. The classification of PL ground motions continues to be a significant challenge due to the absence of consensus on their definition and categorization. This paper investigates the potential benefits of several Machine Learning Classifiers (MLCs) algorithms such as decision tree, random forest, logistic regression, naive Bayes, support vector machine, K-nearest neighbor, ensemble model, and artificial neural network models for predicting PL and Non-Pulse-Like (NPL) ground motions. In this regard, a dataset comprising 200 near-fault ground motions records compiled from active tectonic regions like Taiwan, Turkey, Iran, and Japan, was divided into 2 portions, with 75% used to train the model and the remaining 25% used for testing. Plots of performance metrics, confusion matrix, and receiver operating curve indicate that the ensemble classifier outperforms the other classifier with 86.2% accuracy and the lowest misclassification of PL and NPL ground motions. Additionally, the trained MLC has been compared with the existing ground motion classification models to further assess the accuracy of the different classifiers in the present study. [ABSTRACT FROM AUTHOR]

The secondhand vehicle market in Türkiye has always been active. In the secondhand vehicle market, information such as brand, model, and fuel type can be analyzed, and this information can be made useful. Prices vary according to the various features of the vehicles. Machine learning techniques can predict prices and help users set prices when selling or buying vehicles. Price prediction falls under regression. Since the number of secondhand vehicles is quite high, big data systems are used. Apache Spark and its machine learning library are quite useful for this. Linear regression, decision tree regression, random forest regression, GBT regression, and isotonic regression algorithms are used for price prediction. The random forest algorithm achieved the highest success for the price prediction with 21435.09 RMSE and 0.887 R2 values. As a result of the statistical tests performed to check the significant difference between the RMSE and R2 values obtained with the random forest algorithm and other algorithms, it is concluded that the results obtained with the random forest algorithm are statistically better than other algorithms. The random forest algorithm gives better results because the algorithm performs training over multiple decision trees, its flexibility, and strong hyperparameters. [ABSTRACT FROM AUTHOR]

This study estimates intra-daily PM10 concentrations at 213 inland and coastal monitoring sites in Türkiye from 2008 to 2019 using satellite-based aerosol optical depth (AOD) from the Moderate Resolution Imaging Spectroradiometer (MODIS). An estimation model based on the random forest (RF) approach was developed using the AOD data from the Terra satellite, the meteorological data, and aerosol diagnostics from the Modern-Era Retrospective analysis for Research and Applications, version 2 (MERRA-2), and other auxiliary variables. First, the correlation between the matched PM10 concentrations and MODIS AOD was investigated simply with the quadrant regression (QR) approach. Next, the feature selection procedure was applied to obtain the most significant predictive variables for the estimation model. Then, the spatial and temporal performances of the developed RF model were intensely discussed. Finally, a bias analysis based on the most influential input parameters was also performed to examine the potential errors in the estimated PM10 concentrations. As a result, the RF model showed moderately good performance, with a correlation coefficient (R) of 0.72 and low root mean square error (RMSE) over the entire country, which was better than the results of previous studies in the region. Moreover, the model better estimated PM10 concentrations at individual monitoring sites (with R up to 0.90), particularly in coastal regions. However, overfitting occurred in areas with low populations and few monitoring stations. Additionally, the RF model's performance varied slightly across different seasons, such as autumn (R ≅ 0.69), spring (R ≅ 0.65), winter (R ≅ 0.64), and summer (R ≅ 0.60), and it did not adequately estimate intra-daily PM10 concentrations at the seasonal scale. Furthermore, the bias analyses indicated that higher PM10 and dust mass concentrations, u and v wind components were significant parameters that caused bias in the estimations. Finally, this study provides valuable information for further applications in PM10 patterns and represents the first step toward constructing a high-resolution satellite-based air quality monitoring network in Türkiye. [ABSTRACT FROM AUTHOR]

Yield estimation is an important field of study in agriculture. Forecasting yields provides producers, consumers, traders and policymakers with important preliminary information and time to take necessary action. Corn is an important product in terms of international trade and is widely used in human and animal nutrition throughout the world. Adana produces the highest amount of corn sown both as main and secondary product in Türkiye. Therefore, in this study, corn yield was tried to be estimated by using various meteorological parameters and plant fertilizer usage amounts. For this purpose, statistical (Auto-ARIMA), machine learning (Random Forest) and deep learning (CNN, LSTM) methods were used. The study findings showed that all models used predicted maize yield highly accurately. However, the highest accuracy LSTM model estimated the yield of first corn crop. [ABSTRACT FROM AUTHOR]

Simple Summary: The donkey (Equus asinus) is an odd-toed ungulate and the smallest species in the Equidae family. It is characteristically short-legged with extremely long ears. The wild ancestor of the donkey is equally Equus asinus, which is generally known as the "African wild ass" and is reportedly still extant. Donkeys are the only ungual animal domesticated exclusively in Africa. By nature, donkeys are very companionable, calm, enduring, intelligent, prudent, playful, and keen to learn, and they enjoy the company of humans. In Turkey, donkeys are used for pack transport and riding in order to lessen the physical load on humans. This study was conducted to assess the prediction performance of various algorithms using the morphological traits, body coat colour distribution, and body measurements of donkeys raised in Turkey. This study was carried out in order to determine the morphological characteristics, body coat colour distribution, and body dimensions of donkeys raised in Turkey, as well as to determine the relationships between these factors. For this reason, the predictive performance of various machine learning algorithms (i.e., CHAID, Random Forest, ALM, MARS, and Bagging MARS) were compared, utilising the biometric data of donkeys. In particular, mean measurements were taken from a total of 371 donkeys (252 male and 119 female) with descriptive statistical values as follows: height at withers, 100.7 cm; rump height, 103.1 cm; body length, 103.8 cm; chest circumference, 112.8 cm; chest depth, 45.7 cm; chest width, 29.1 cm; front shin circumference, 13.5 cm; head length, 55 cm; and ear length, 22 cm. The body colour distribution of the donkeys considered in this study was calculated as 39.35% grey, 19.95% white, 21.83% black, and 18.87% brown. Model fit statistics, including the coefficient of determination (R2), mean square error, root-mean-square error (RMSE), mean absolute percentage error (MAPE), and standard deviation ratio (SD ratio), were calculated to measure the predictive ability of the fitted models. The MARS algorithm was found to be the best model for defining the body length of donkeys, with the highest R2 value (0.916) and the lowest RMSE, MAPE, and SD ratio values (2.173, 1.615, and 0.291, respectively). The experimental results indicate that the most suitable model is the MARS algorithm, which provides a good alternative to other data mining algorithms for predicting the body length of donkeys. [ABSTRACT FROM AUTHOR]

Rampant terrorism poses a serious threat to the national security of many countries worldwide, particularly due to separatism and extreme nationalism. This paper focuses on the development and application of a temporal self-exciting point process model to the terror data of three countries: the US, Turkey, and the Philippines. To account for occurrences with the same time-stamp, this paper introduces the order mark and reward term in parameter selection. The reward term considers the triggering effect between events in the same time-stamp but different order. Additionally, this paper provides comparisons between the self-exciting models generated by day-based and month-based arrival times. Another highlight of this paper is the development of a model to predict the number of terror events using a combination of simulation and machine learning, specifically the random forest method, to achieve better predictions. This research offers an insightful approach to discover terror event patterns and forecast future occurrences of terror events, which may have practical application towards national security strategies. [ABSTRACT FROM AUTHOR]

The evolution of ensemble learning has recently offered a new approach to model complex systems. Inspired by the success of such methods, this paper introduces a new ensemble approach that integrates capabilities of two top state-of-the-art machine learning (ML) methods, namely random forests (RF) and genetic programming (GP), to model and forecast meteorological drought onset and severity. The new method, called gene-random forest (GeRF), follows the same steps of a standard GP systems, but with differences in the generation of initial population of potential solutions. The GeRF was tested to model and predict standardized precipitation evapotranspiration indices (SPEI-3 and SPEI-6) at two meteorology stations in Ankara province, Turkey. We have compared its efficiency to those of a classic autoregressive model as well as standalone RF, GP, and a hybrid ML model, called Bat-ELM, achieving results meaningfully superior to the benchmarks, particularly in the testing data. [ABSTRACT FROM AUTHOR]

Near-surface air temperature is a key variable used in a wide range of applications showing the environmental conditions of the Earth. Standard meteorological observations generally provide the best estimation with high accuracy over time for a small area of influence. However, considerable uncertainty arises when point measurements are extrapolated or interpolated over much larger areas. Satellite and reanalysis products have emerged as a viable alternative or supplement to in situ observations due to their availability over vast ungauged regions. Thus, spatial patterns of air temperature can be derived from these products. In this study, we evaluate the performance of several reanalysis products of near-surface air temperature to determine the best product in estimating daily and monthly air temperatures across the complex terrain of Turkey. European Center for Medium-Range Weather Forecasts Reanalysis version 5 (ERA5), (AgERA5), (ERA5-Land), the Modern-Era Retrospective Analysis for Research and Applications version 2 (MERRA2), and the Japanese 55-year Reanalysis (JRA-55) products are assessed with 1120 ground-based gauge stations for the period 2015–2019 over complex terrain having different climate classes according to Köppen-Geiger classification scheme and land surface types. Moreover, several traditional and more sophisticated machine learning downscaling algorithms are applied to increase the products' spatial resolution. The agreement between ground observations and the different products and the downscaled temperature product is investigated by using a set of commonly employed statistical estimators of mean absolute error (MAE), correlation coefficient (CC), root-mean-square error (RMSE), and bias. Performance analysis of reanalysis air temperature products with ground-based observations on daily and monthly time series has shown that among the five datasets, the AgERA5 product is superior in estimating air temperature over Turkey, both seasonally and annually. Compared to the other datasets, ERA5 and ERA5-Land show similar performance and reach the second highest performance score at both daily and monthly time steps. In terms of improving product performance with spatial downscaling, among the distance-based methods, the best overall performance is obtained by bicubic interpolation with a slight increase in the product performance in monthly and daily time series. However, depending on the season, the Random Forest algorithm's performance is far superior to all other methods used in this study. [ABSTRACT FROM AUTHOR]

Evapotranspiration (ET), as one of the most central parameters to the climate systems, plays a predominant role in the water, energy, carbon cycles, and energy–moisture exchanges between the earth and the atmosphere. Since the in-situ estimation of ET is challenging, the remote sensing-based techniques are deemed a reliable surrogate for the traditional ET estimation methods. The Gravity Recovery And Climate Experiment (GRACE) mission turned out to perform well in detecting the variations of ET over large basins. However, the subbasin-wise variations of ET based on the GRACE mission remain challenging due to its coarse resolution. In this study, the GRACE/GRACE-FO observations were downscaled based on Random Forest Machine Learning (RFML) algorithm and were integrated with fine-resolution precipitation and runoff data within the framework of the terrestrial water balance approach to derive small-scale variations of ET over Kizilirmak Basin (KB) in Türkiye. The results were compared to the fine-resolution ET products of the Moderate Resolution Imaging Spectroradiometer (MODIS) satellite and Noah model and revealed acceptable precision of the methodology for the subbasin-wise application. The findings suggest that excluding the Lower Kizlirmak Basin (LKB), the water balance-based ET manifests a very high agreement with the MODIS- and Noah-derived ET over all the subbasins of the KB. The impacts of the main ET driving parameters such as temperature, water availability, solar radiation, surface albedo, and vegetation status were further investigated in terms of linear correlations. It is found that there is a high association between the monthly variations of ET and its conditioning factors over the KB, especially between the water availability and ET. [ABSTRACT FROM AUTHOR]

Legumes are the most basic carbohydrate, fiber and protein sources and provide essential vitamins and minerals for a healthy nutrition. As a gene center for legumes, edible legumes are the most cultivated field crops after cereals in Turkey and mainly produced ones are chickpeas, green and red lentils, beans and broad beans. Besides, as the agricultural production is very sensitive to weather conditions, it makes the current global climate change even more important because of its negative effects of on the environment, natural resources and production systems. In this study, by using a data set including meteorological factors as average humidity, average temperature, average precipitation in addition to cultivation and production values, estimations of production amounts and yields were performed for chickpeas and lentils by using Support Vector Regression (SVR), Decision Tree Regression (DTR) and Random Forest Regression (RFR) machine learning methods for 2021 and 2022. Also, accuracy and validity of the methods were evaluated by comparing obtained predictions with actual values and by considering various performance evaluation criteria. According to the obtained results by all three methods, in 2022 a 25% decrease in chickpea production and an increase of 25-65% in lentil production is predicted compared to the previous year. In addition, it can also be stated that tree-based machine learning methods yielded more successful outcomes in both predictions. [ABSTRACT FROM AUTHOR]

The increasing urbanization and rapid economic development have led to growing housing markets. This has resulted in a rising demand for housing due to migration towards cities and the acceptance of housing as both an investment and a wealth protection tool. On the other hand, housing prices are affected by many factors, thereby studies on housing price forecasts are carried out within the scope of different models and various methods. In line with the significant importance of housing markets as one of Turkey's economic growth tools, this study includes the forecast of the classification of 4-bedroom (3+1) housing prices in the central district of Kütahya province. The dataset used includes various criteria that affect housing prices, and different machine learning classification algorithms that are utilized. As a result, it is seen that the accuracy rates of the k-Nearest Neighbor (k-NN), Decision Trees (DT) and Random Forest (RF) algorithms were above 60%. As the main performance criterion, the AUC scores yielded that the most successful classification forecast is acquired by the RF algorithm, which is followed by the k-NN. In the context of the variables and data set included in this study, it was seen that successful outcomes in the classification forecasts of 4-bedroom apartments’ housing prices in the city center of Kütahya were obtained by these two methods. [ABSTRACT FROM AUTHOR]

Occupational accidents are frequent in the construction industry, containing significant risks in the working environment. Therefore, early designation, taking preventive actions, and developing a proactive safety risk management plan are of paramount significance in managing safety issues in the construction industry. This study aims to develop a national data-driven safety management framework based on accident outcome prediction, which helps anatomize precursors of fatalities and thereby minimizing fatal accidents on construction sites. A national data set comprising 338,173 occupational accidents recorded in the construction industry across Turkey was used to develop a data-driven model. The random forest algorithm coupled with particle swarm optimization was used for the prediction and the interpretability of the proposed model was augmented through the game theory–based Shapley additive explanations (SHAP) approach. The findings showed that the proposed algorithm achieved satisfactory model performances for detecting construction workers who might face a fatality risk. The SHAP analysis results indicated that both company (such as number of past accidents and workers in the company) and worker-related (such as age, daily wage, experience, shift, and past accident of the workers) attributes were influential in identifying fatalities by detecting which workers might face fatal accidents under which conditions. A construction safety management plan was developed based on the analysis results, which can be used on construction sites to detect workers/conditions that are most susceptible to fatalities. The findings of the present research are expected to contribute to orchestrating effective safety management practices in construction sites by characterizing the root causes of severe accidents. [ABSTRACT FROM AUTHOR]

Various machine learning (ML) techniques have been recommended and used in the literature to produce landslide susceptibility map (LSM). On the other hand, feature engineering (FE) is an important topic in ML studies, but the concept is ignored by most research. In this study, a novel FE framework, including feature selection, feature transformation, feature binning, and feature weighting, is proposed to produce LSMs using eXtreme gradient boosting (XGBoost), random forest (RF), and support vector machine (SVM). For this purpose, first, thirteen landslide conditioning factors used in data preprocessing were utilized for producing LSM models in the study area, Babadag district of Denizli Province in the Aegean region of Turkey. Second, two irrelevant factors eliminated from the input feature subset using the feature selection in the FE framework. Third, features determined as skewed data were converted into symmetric form by applying feature transformation analysis with log transformation. Then, the remaining factors having continuous values were turned into categorical values using the quantile classifier technique. During the feature weighting phase, four different feature weighting methods, namely, eXtreme Gradient Boosting, random forest (RF), non-negative least squares (NNLS), and Frequency Ratio, were utilized to calculate the weights in each subclass of each landslide-related factor. In addition, the proposed feature subsets were also compared with raw data. At the end of process, the XGBoost model constructed with a FR-selected subset (Overall Accuracy (Acc) = 0.907 and area under curve (AUC) = 0.9822) outperformed both raw (Acc = 0.874; AUC = 0.960) and other methods (i.e., RF–FR and SVM–NNLS). Consequently, the study results revealed that the proposed FE approach could be a useful framework to increase the performance of ML techniques in identifying and extracting relevant features to develop highly optimized and enriched models. [ABSTRACT FROM AUTHOR]

This study, which aims to make predictions using a previously unused deep hybrid Convolutional Neural Network-Support Vector Regression approach for hydropower generation, was carried out using a unique meteorological data set as input data. The unique dataset was collected from Kaman Meteorology Directorate and Hirfanlı HEPP in Turkey, from 2007 to 2021, to estimate the HEPP's Net Head and Hydroelectric Power Generation on a daily basis. The performances of the prediction models were benchmarked in addition to the used CNN-SVR model with Machine Learning (ML) models (Boosting Random Forest Regression (BRFR) and Weighted K-Nearest Neighbor Regression (WKNNR)) and Deep Learning (DL) models (Long-Short Term Memory (LSTM) and Deep Belief Network (DBN)). The comparison results of the used CNN-SVR model with other alternative models showed that CNN-SVR performed effectively with the highest correlation coefficient of 0.971 for NH and 0.968 for PP. [ABSTRACT FROM AUTHOR]

This research introduces a scientific methodology to identify areas affected by gully erosion using Geographic Object Based Image Analysis (GEOBIA) and Random Forest (RF) supervised machine learning. The GEOBIA and RF were applied in Besni district, which has a Mediterranean climate, of Adiyaman province in Turkey by including many factors in the model. Estimation Scale Parameter (ESPII) algorithm was used in the segmentation phase. The novelty of this study is the implementation of RF supervised classification algorithm to classify a large number of objects determined after the segmentation process, due to the large size of the study area. Therefore, open access data has been evaluated with high classification accuracy without the need for labor. Precision, Recall and F1-Score values were calculated using true positive (TP), true negative (TN), false positive (FP) and false negative (FN) values based on field observations and Google Earth images of the study area. The TP, TN, FP and FN values were 0.90, 0.95 and 0.92, respectively. In addition, a Kappa-index was calculated as 0.88. The gully erosion map obtained using aforementioned methodology can be used to take necessary measures to prevent further degradation and plan sustainable land uses. [ABSTRACT FROM AUTHOR]

Predicting medical waste (MW) properly is vital for an effective waste management system (WMS), but it is difficult because of inadequate data and various factors that impact MW. This study's primary objective was to develop an ensemble voting regression algorithm based on machine learning (ML) algorithms such as random forests (RFs), gradient boosting machines (GBMs), and adaptive boosting (AdaBoost) to predict the MW for Istanbul, the largest city in Turkey. This was the first study to use ML algorithms to predict MW, to our knowledge. First, three ML algorithms were developed based on official data. To compare their performances, performance measures such as mean absolute deviation (MAE), root mean squared error (RMSE), mean absolute percentage error (MAPE), and coefficient of determination (R-squared) were calculated. Among the standalone ML models, RF achieved the best performance. Then, these base models were used to construct the proposed ensemble voting regression (VR) model utilizing weighted averages according to the base models' performances. The proposed model outperformed three baseline models, with the lowest RMSE (843.70). This study gives an effective tool to practitioners and decision-makers for planning and constructing medical waste management systems by predicting the MW quantity. [ABSTRACT FROM AUTHOR]

Mountain vipers of the genus Montivipera, generally, and the species of the Raddei clade, specifically, are interesting examples of species neo-endemism in Iran, Anatolia, and the Caucasus. Given the critical conservation status of these species, it is necessary to identify their suitable habitats for prioritizing conservation measures. We modeled ecological niche of each species based on four species distribution models, including generalized linear models, generalized boosting models, random forest, and maximum entropy and combined them in an ensemble model. Also, using a new principal component analysis (PCA-env), the ecological niche divergence of the species was investigated. Results indicated that all models with AUC and TSS > 0.9 had excellent predictive performance. The species of the Raddei clade showed different patterns of ecological niche occupation and the greatest differentiation was seen in the gradient of temperature seasonality, annual precipitation, and topographical ruggedness. M. latifii and M. kuhrangica, despite the great geographical distance, revealed the highest degree of niche overlap and niche similarity. Due to the restricted distribution and limited evolutionary adaptability of mountain vipers in the mountainous regions of Alborz and Zagros, it is imperative to plan conservation measures to reduce the factors that threaten the long-term survival of these species. [ABSTRACT FROM AUTHOR]

Across the globe, landslides have been recognized as one of the most detrimental geological calamities, especially in hilly terrains. However, the correct determination of landslide-prone fields remained a challenging task due to the nonlinear, complex, and inconsistent nature of landslides. Therefore, it is essential to apply methods with superior capabilities in dealing with real-world problems for properly delineating potential landslide zones. Recently, ensemble learning techniques have been drawn intense interest in landslide susceptibility mapping studies due to their distinct advantages. This present work intended to propose natural gradient boosting (NGBoost), a novel member of the ensemble learning family, for modeling landslide susceptibility for Macka County of Trabzon province, Turkey. The predictive performance of NGBoost was compared to ensemble-based machine learning methods, namely random forest (RF) and XGBoost using five accuracy metrics including overall accuracy (OA), F1 score, Kappa coefficient, area under curve (AUC) value, and root-mean-square error to evaluate its competence and robustness. Besides, SHAP based on the game theory approach was implemented to interpret the influences of the predisposing factors on the produced model. Results indicated that the NGBoost method utilized for landslide susceptibility mapping problem for the first time had the greatest predictive ability (AUC = 0.898), followed by XGBoost (AUC = 0.871) and RF (AUC = 0.863), and outperformed the XGBoost and RF by approximately 6% in terms of OA. McNemar's statistical significance test results also confirmed the superiority of the proposed NGBoost method over the RF and XGBoost algorithms. [ABSTRACT FROM AUTHOR]

• Burn severity mapping was conducted with dNBR and ensemble-based classifiers. • Machine learning methods were used using pixel- and object-based image analyses. • Canonical correlation forest outperformed RF and RotFor by about 5%. • USGS FIREMON and Mediterranean condition dNBRs were highly correlated. Forest fires cause aerosol emissions and biomass burning that pose major threats to the ozone layer. The precise estimation of burned area with the degree of burn severity plays a critical role to investigate the impacts of fire on forests. Burn severity analysis using remotely sensed data has become popular in the last decade for post-fire detection, management and mitigation studies. In this study, performances of three ensemble learning algorithms, namely, random forest (RF), rotation forest (RotFor), and canonical correlation forest (CCF) were analyzed on burn severity mapping in terms of pixel-and object-based image analyses applying standard accuracy metrics and McNemar's statistical test. The classification procedures were applied to two fire-affected areas located in Izmir province, Turkey using multitemporal (pre-and post-fire) Sentinel-2A images. The results showed that the CCF classifier outperformed the RF and RotFor classifiers for both pixel-and object-based classifications in that classification accuracies varied by about 5% in both pixel-and object-based classifications. According to the results of pixel-based CCF classification, total burned area (i.e., moderate-high and high severity) was estimated as 813 ha and 2,546 ha for Menderes and Karabağlar fires, respectively. The difference between the pixel- and object-based classification estimation about the burned area coverage could be resulting from intraclass spectral heterogeneity about the image objects located in the burned regions of the study areas. Statistical analysis also revealed that the RotFor and CCF algorithms produced similar results in pixel-based and object-based classification for both fires. [ABSTRACT FROM AUTHOR]

The aim of this study is to examine the capability of the combined LiDAR/WorldView-3 data in estimating the plot-level stand attributes (stem number-N, mean diameter-D, mean height-H, basal area-BA and volume-V) in a complex forest located in the northwest of Turkey. Total 135 plots were measured to determine the forest attributes. Prediction models were developed at three levels which are: i) the general level for all stands (including all plots), ii) forest type level (coniferous forest, broad-leaved forest), and iii) tree species level (Black pine stands, Maritime pine stands, Oak stands, Mixed stands). Multiple Linear Regression (MLR) and Random Forest (RF) modelling approaches were tested to predict stand attributes. The MLR regression modelling showed that the stand attributes were estimated with R2 ranging from 0.71 (N and V in Mixed) to 0.94 (H in Maritime pine) at tree species level, from 0.73 (BA in Broadleaved) to 0.95 (H in Conifer) at forest types level and from 0.77 (V) to 0.89 (H) at general level. The RF modelling indicated that the stand attributes were estimated with R2 ranging from 0.69 (V in Mixed and Oak) to 0.94 (H in Maritime pine) at tree species level, from 0.72 (N in Broadleaved) to 0.95 (H in Conifer) at forest types level and from 0.81 (N and V) to 0.88 (D) at general level. The mean height had the highest prediction accuracy for almost all levels in both approaches. However, the stem number and basal area were generally estimated with the lower accuracies. The homogeneous coniferous stands provided the higher estimation accuracy than the broadleaved stands. Our results showed that the modelling approaches used here provide different performance for predicting different stand attributes. While the MLR approach performed better in estimating the stand attributes at the tree species level, the RF approach towards the general level provided higher accuracy estimation. In conclusion, the combination of aerial laser scanning and high resolution satellite data has high potential for predicting stand attributes in complex forest ecosystems. [ABSTRACT FROM AUTHOR]

Objective: In this study, the number of bacteriae in the air at five different regions in Eskişehir (Turkey) according to the settle plate method and some meteorological factors such as temperature and relative humidity were collected and analyzed statistically. The data were used to predict the amount of bacteriae using machine learning algorithms. Material and Methods: Air samples were taken from specified sampling points using the settle plate method. The random forest method, multivariate adaptive regression splines, support vector machine and ordinary least squares models were employed to analyze a qualified data set collected through 5 different monitoring locations in Eskişehir city, Turkey. All analyses were conducted using RStudio Software (version 1.2.1335). Results: The results revealed that all variables including weather characteristics and particulate matters were found statistically significant on the amount of the bacteria. Among modeling methods, the random forest model overperformed the others in terms of root mean square. Conclusion: Results might be useful for researchers studying bacterial density. These findings may provide policy-makers with crucial information for a better climate and health policy development, which helps address the conflict between development and air pollution, may be applicable in Turkey and other countries as well. [ABSTRACT FROM AUTHOR]

A gradient boosting regression tree (GBT) approach is introduced for 1- and 3-month-ahead standardized precipitation–evapotranspiration index (SPEI) classification for Antalya and Ankara in Turkey. First, the numerical target series of SPEI-6 was converted into the categorical vectors of extreme wet, wet, near normal, dry, and extremely dry labels. Then, a GBT model was trained and validated using the lagged SPEI-6 series of four global grid points closest to each city. The model efficiency was surveyed in terms of kappa, overall accuracy, misclassification rate, class recall, class precision, and F1 score. The GBT was also compared with the traditional decision tree and state-of-the-art random forest models developed as the benchmarks. Despite facing the unbalanced label vectors, the GBT approach showed promising processing performance with an overall accuracy of 77% (83%) and 74% (72%) in Antalya (Ankara) at 1- and 3-month-ahead prediction scenarios, respectively. [ABSTRACT FROM AUTHOR]

Landslide susceptibility mapping (LSM) is a major area of interest within the field of disaster risk management that involves planning and decision-making activities. Therefore, preparation of dataset, construction of predictive model and analysis of results are considered to be important stages for effective and efficient disaster management in LSM. In recent years, a large number of studies has mainly focused on the effects of using machine learning (ML) algorithms as a predictive model in LSM. Decision tree-based ensemble learning algorithms known as decision forest is one of the popular ML techniques based on a combination of several decision tree algorithms to construct an optimal prediction model. In this study, prediction performances of recently proposed decision tree-based ensemble-based algorithms namely canonical correlation forest (CCF) and rotation forest (RotFor) are tested on LSM. In order to compare their performances, popular ensemble learning algorithms including random forest (RF), AdaBoost and bagging algorithms are also considered. For this purpose, first, twelve conditioning factors are determined in the study area, Karabuk province of Turkey. Second, individual importance of the factors on LSM process is evaluated using Fischer score analysis and selected factors are used as an input dataset for the construction of landslide susceptibility prediction models of CCF, RotFor, RF, AdaBoost and bagging algorithms. For the assessment of the performances, overall accuracy (OA), success rate curves and the area under the curve (AUC) analysis are utilized. Furthermore, chi-squared-based McNemar's test and well-known accuracy measures known as receiver operating characteristic (ROC) curves are employed to evaluate the pairwise comparison of the ensemble learning methods. Results show that CCF method outperforms the RotFor method by about 4%, and there is no statistically significant difference between CFF and other methods. [ABSTRACT FROM AUTHOR]

Image analysis techniques are developing as applicable to the approaches of quantitative analysis, which is aimed to determine cultivar grains. Additionally, corn (Zea mays) grain processing companies evaluate the quality of kernels to determine the price of these cultivars. Because of this reason, in the study, a computer image analysis technique was applied on three corn cultivars. These were Zea mays L. indentata, Zea mays L. saccharata and a hybrid corn (Yellow sweet corn). These cultivars are commercially important as dry grains in Turkey. In the study, the grain color values were tested in the cultivars from Turkey's collection. One hundred samples were used for each corn cultivar, and 300 corn grains in total were used for evaluations. Each of nine color parameters (Rmin, Rmean, Rmax, Gmin, Gmean, Gmax, Bmin, Bmean, Bmax) which were obtained from original RGB color channels with maximum and minimum values was evaluated from the digital images of three different corn cultivar grains. The values were analyzed with the help of the Multilayer Perceptron (MLP), Decision Tree (DT), Gradient Boost Decision Tree (GBDT) and Random Forest (RF) algorithms by using the Knime Analytics Platform. The majority voting method was applied to MLP and DT for prediction fusion. All algorithms were run with a 10-fold cross-validation method. The success of prediction accuracy was found as 99% for RF and GBDT, 97.66% for MLP, 96.66% DT and 97.40% for Majority Voting (MAVL). The MAVL method increased the accuracy of DT while decreasing the accuracy of MLP partly for the fusion of MLP and DT. [ABSTRACT FROM AUTHOR]

This paper presents a new tree-based model, namely Fuzzy Random Forest (FRF), for one month ahead Standardized Precipitation Evapotranspiration Index (SPEI) classification and prediction with a noteworthy application in ungauged catchments. The proposed FRF model uses global SPEI dataset as the meteorological drought quantifier and applies a fuzzy inference system to extract fuzzified and crisp SPEI values for an ungauged catchment. The evolved crisp series is then transformed into the polynomial label vector of extremely wet, wet, near normal, dry, and extremely dry categories. In the end, the state-of-the-art random forest algorithm was used to classify and predict the label vector using the lagged SPEI series of the selected global grid points. To demonstrate the development and verification process of the FRF model, the global SPEI-6 values for the period of 1961–2015 were retrieved from four global grid points around the Central Antalya Basin, Turkey. The new model was trained and validated using 70% and 30% of the data sets, respectively. The performance of the new model was examined in terms of total accuracy, misclassification, and Kappa statistics and cross-validated with the fuzzy decision tree model developed as the benchmark in this study. The results showed the promising performance of the FRF for drought classification and prediction with outstanding efficiency for extremely wet and dry events classification. According to the Kappa statistic, the proposed FRF model is 25% more accurate than the benchmark FDT model. [ABSTRACT FROM AUTHOR]

The motion picture industry is one of the largest industries worldwide and has significant importance in the global economy. Considering the high stakes and high risks in the industry, forecast models and decision support systems are gaining importance. Several attempts have been made to estimate the theatrical performance of a movie before or at the early stages of its release. Nevertheless, these models are mostly used for predicting domestic performances and the industry still struggles to predict box office performances in overseas markets. In this study, the aim is to design a forecast model using different machine learning algorithms to estimate the theatrical success of US movies in Turkey. From various sources, a dataset of 1559 movies is constructed. Firstly, independent variables are grouped as pre-release, distributor type, and international distribution based on their characteristic. The number of attendances is discretized into three classes. Four popular machine learning algorithms, artificial neural networks, decision tree regression and gradient boosting tree and random forest are employed, and the impact of each group is observed by compared by the performance models. Then the number of target classes is increased into five and eight and results are compared with the previously developed models in the literature. [ABSTRACT FROM AUTHOR]

In this study, two kernel-based models were used which include Support Vector Regression (SVR) and Gaussian Process Regression (GPR) and were compared with two tree-based models that are M5 and Random Forest (RF) for estimating missing monthly precipitation data in Antakya, Dortyol, Iskenderun and Samandag stations, which are the important precipitation stations in the Eastern Mediterranean region, Turkey. For this purpose, firstly 10% random precipitation data were assumed as missing data for the period 1980-2019. Secondly, the missing data in each station was estimated with the data of other stations within the framework of four data combinations scenarios. In Kernel-based SVR and GPR methods, the RBF kernel gave suitable results for the selected study area. While SVR and RF methods gave very close estimation results, the SVR method gave relatively better results than the other methods especially in error minimizing aspects. Gaussian function based GPR model generally tries to estimate missing data closer to means. This is the main disadvantage of the GPR model and therefore it is unsuccessful in the estimation process. Finally, the results showed that the algorithms based on machine learning are successful in estimating the missing precipitation data. [ABSTRACT FROM AUTHOR]

• Random forest machine learning algorithm perform very well in simulating the subbasin scale variations of Terrestrial water storage. • The integrated analysis of the GRACE and fine resolution hydrological modelling, turns out to be a promising methodology for small scale assessment of flooding. • The Kizilirmak Basin (KB) has been susceptible to extreme hydrological event of inundation in Feb 2013. • Soil Moisture and Precipitation are two dominant flood-conditioning parameters of the KB. The Gravity Recovery and Climate Experiment (GRACE) paved the way for large-scale monitoring of the hydrological extremes. However, local scale analysis is aslo challenging due to the coarse resolution of the GRACE estimates. The feasibility of the downscaled GRACE data for the flood monitoring in the Kizilirmak Basin (KB) in Türkiye is investigated in this study by integrating the GRACE and hydrological model outputs of a random forest approach. Results suggest that the TWSA, over the Asagi Kizilirmak Basin (AKB), is ascending with an annual rate of + 3.51 m m / y r ; while the Orta Kizilirmak Basin (OKB), Yukari Kizilirmak Basin (YKB), Delice Basin (DB), Develi Kapali Basin (DKB), and Seyfe Kapali Basin (SKB) showed descending trend respectively as - 1.15 m m / y r , - 1.58 m m / y r , - 1.14 m m / y r , - 2.34 m m / y r , and - 1.31 m m / y r. The hydrological status of the basin showed that in 2003, 2005, 2010–2013, and 2015–2016 periods the study area was prone to the inundation. Hence, by validating the Flood Potential Index (FPI) rates acquired from the downscaled GRACE data, it was shown that the best correlation coefficient (0.73) between FPI and streamflow (Q) is associated with the SKB. It is also concluded that the downscaled TWSA associated with the fine-resolution models depicts acceptable accuracy in determination of the flood potential at local scales. [ABSTRACT FROM AUTHOR]

• Machine learning models can detect injured body parts of workers. • A novel inclusive multi-stage ensemble classification scheme is introduced. • Class imbalance issue is handled using several resampling techniques. • Proposed approach improved the prediction performance significantly. • Random forest algorithm outperforms other methods in classification. Despite significant progress has been made in safety management practices, construction industry still accounts for a substantial number of occupational accidents leading to injuries in different body parts of construction workers. In this vein, detecting the most susceptible body parts of construction workers by using only pre-accident information is of particular importance by means of performance augmentation using advanced modelling techniques as it helps safety managers orchestrate the most relevant and adequate mitigation measures. The central focus of this study is to identify the susceptible body parts of construction workers proactively and propose measures specific to the corresponding body parts. Hence, this study aims to develop a machine learning (ML)-based novel inclusive multi-stage ensemble model that identifies the most vulnerable body parts of construction workers using a national dataset recorded in Turkey. Findings illustrate that incorporating ensemble modelling approach into predictions enhanced accuracies and the ensemble random forest (RF) model reinforced with principal component analysis (PCA) yielded the best performance. Results further highlight that number of workers in the company, working days of the worker, and age of the worker are the most influential attributes in the susceptibility of body parts. A utilization plan is developed based on the analysis results, which can be run monthly on construction sites to identify the most vulnerable body parts of construction workers. Overall, this study is expected to contribute to the development of more robust safety management applications by allowing safety managers to evaluate susceptible body parts of construction workers prior to accidents. [ABSTRACT FROM AUTHOR]

The mastery in forecasting the streamflow rates is of great importance in the design, planning and resilience against droughts. Likewise, the application of flow duration and mass flow curves in the design of the reservoir capacity, energy generation, water allocation, etc. especially at the tributary reaches is a great challenge mostly due to the lack of information and data records. In this study, we have developed a methodology to obtain the flow duration curve (FDC) and mass flow curve (MFC) in tributary stream stations with the help of estimated streamflow rates. The procedure suggests using two alternative approaches in the selection of the reference station on the mainstream. The streamflow in the reference station is decomposed into direct runoff (DR) and base flow (BF) using one-parameter digital filter method. Together with the precipitation records in the tributary station, the DR and BF on the reference station are then used to estimate the FDC and MFC. The multivariate adaptive regression spline (MARS) and random forest (RF) methods are used to alternate each other, and the residual of the models are simulated using the autoregressive conditionally heteroscedastic (ARCH) approach to develop the hybrid MARS-ARCH and RF-ARCH models. A data set related to Coruh River Basin, in Turkey is used to confirm the methodology, while results with R 2 ≥ 0.92, reasonable bias, and relative error in the estimation of the expected FDC and MFC rates indicated the robustness of the suggested methodology. [ABSTRACT FROM AUTHOR]

• The form of the forecast equation for factor models is constructed by ML models. • RF delivers regularly consistent performance without using a factor model. • Tree-based models exhibit superior performance with a high number of predictors. • Explainable inflation forecasts are produced by ML models. Forecasting inflation accurately in a data-rich environment is a challenging task and an active research field which still contains various unanswered methodological questions. One of them is how to find and extract the information with the most predictive power for a variable of interest when there are many highly correlated predictors, as in the inflation forecasting problem. Traditionally, factor models have been used to tackle this problem. However, a few recent studies have revealed that machine learning (ML) models such as random forests may offer some valuable solutions to the problem. This study encourages greater use of ML models with or without factor models by replacing the functional form of the forecast equation in a factor model with ML models or directly employing them with several feature selection techniques. This study adds new tree-based models to the analysis in the light of the recent findings in the literature. Moreover, it proposes the integration of feature selection techniques with Shapley values to find out concise explanations of the inflation predictions. The results obtained by a comprehensive set of experiments in an emerging country, Turkey, facing a high degree of volatility and uncertainty, indicate that tree-based ensemble models can be advantageous by providing better accuracy together with explainable predictions. [ABSTRACT FROM AUTHOR]

Hazelnuts and tea are two major agricultural crops grown in the eastern Black Sea region in Turkey. Since this part of Turkey is not industrialized, most of the local people work in agriculture, making hazelnuts and tea a part of their lives. For the government side, it is crucial to keep records of the amount of harvested croplands to implement agricultural policies. In fact, the harvested area and crop type of each cadastral parcel are collected either during cadastral surveys or with the declaration of individual farmers, yet this information is mostly not up-to-date and does not reflect the current land-use status. This study aims to determine the extent and distribution of hazelnuts and tea grown areas using the Random Forest (RF) classification algorithm. Tea and hazelnuts give similar spectral reflectance values to surrounding vegetation, which makes it difficult to distinguish them using only their spectral properties. To tackle this problem, the normalized difference vegetation index (NDVI) and texture extraction methods such as the Grey Level Co-occurrence Matrix (GLCM) and Gabor filter were integrated with the RF algorithm, and their contributions to the success of the RF classification method were examined. WorldView-2 satellite images, which have eight multispectral bands (MS: 2 m) and one higher spatial resolution panchromatic band (PAN: 0.5 m), were used. Since the study area contains agricultural products grown in different seasons, satellite images belonging to both summer and winter periods were used. Preliminary results acquired using only spectral values indicated that the RF method gives 79.05% and 71.84% overall accuracies for summer and winter periods, respectively. Integrating texture information improves the performance of the RF algorithm such that the overall classification accuracies are increased to 83.54% and 87.89% when texture information extracted with GLCM and the Gabor filter is added. The classification performance of the winter image is also boosted to be 77.41% and 79.73% with the contribution of texture information obtained with GLCM and the Gabor filter, respectively. Finally, produced thematic maps were compared with the latest cadastral maps to validate classification results with ground truth data. The obtained results reveal the success of integrating texture features in classification since the created thematic maps are consistent with actual land use. The results also show that the crops grown on some cadastral parcels are not coherent with the most current cadastral database, which implies that the cadastral maps need to be updated. [ABSTRACT FROM AUTHOR]

• We improve the success of default prediction models by using graph theory. • Big data platform is used to ensure high performance in terms of time and cost. • Experimental analysis is conducted with more than one million Turkish companies. • The contributions of inputs are demystified by interpretable machine learning. With the unprecedented increase in data all over the world, financial sector such as companies and industries try to remain competitive by transforming themselves into data-driven organizations. By analyzing a huge amount of financial data, companies are able to obtain valuable information to determine their strategic plans such as risk control, crisis management, or growth management. However, as the amount of data increase dramatically, traditional data analytic platforms confront with storing, managing, and analyzing difficulties. Emerging Big Data Analytics (BDA) overcome these problems by providing decentralized and distributed processing. In this study, we propose two new models for default prediction. In the first model, called DPModel-1, statistical (logistic regression), and machine learning methods (decision tree, random forest, gradient boosting) are employed to predict company default. Derived from the first model, we propose DPModel-2 based on graph theory. DPModel-2 also comprises new variables obtained from the trading interactions of companies. In both models, grid search optimization and SHapley Additive exPlanations (SHAP) value are utilized in order to determine the best hyperparameters and make the models interpretable, respectively. By leveraging balance sheet, credit, and invoice datasets, default prediction is realized for about one million companies in Turkey between the years 2010–2018. The default rates of companies range between 3%-6% by year. The experimental results are conducted on a BDA platform. According to the DPModel-1 results, the highest AUC score is ensured by random forest with 0.87. In addition, the results are improved for each technique separately by adjusting new variables with graph theory. According to DPModel-2 results, the best AUC score is achieved by random forest with 0.89. [ABSTRACT FROM AUTHOR]

In the last few years, several countries have accomplished their determined renewable energy targets to achieve their future energy requirements with the foremost aim to encourage sustainable growth with reduced emissions, mainly through the implementation of wind and solar energy. In the present study, we propose and compare five optimized robust regression machine learning methods, namely, random forest, gradient boosting machine (GBM), k-nearest neighbor (kNN), decision-tree, and extra tree regression, which are applied to improve the forecasting accuracy of short-term wind energy generation in the Turkish wind farms, situated in the west of Turkey, on the basis of a historic data of the wind speed and direction. Polar diagrams are plotted and the impacts of input variables such as the wind speed and direction on the wind energy generation are examined. Scatter curves depicting relationships between the wind speed and the produced turbine power are plotted for all of the methods and the predicted average wind power is compared with the real average power from the turbine with the help of the plotted error curves. The results demonstrate the superior forecasting performance of the algorithm incorporating gradient boosting machine regression. [ABSTRACT FROM AUTHOR]

Urban areas may be affected by multiple hazards, and integrated hazard susceptibility maps are needed for suitable site selection and planning. Furthermore, geological–geotechnical parameters, construction costs, and the spatial distribution of existing infrastructure should be taken into account for this purpose. Up-to-date land-use and land-cover (LULC) maps, as well as natural hazard susceptibility maps, can be frequently obtained from high-resolution satellite sensors. In this study, an integrated hazard susceptibility assessment was performed for a developing urban settlement (Mamak District of Ankara City, Turkey) considering landslide and flood potential. The flood susceptibility map of Ankara City was produced in a previous study using modified analytical hierarchical process (M-AHP) approach. The landslide susceptibility map was produced using the logistic regression technique in this study. Sentinel-2 images were employed for generating LULC data with the random forest classification method. Topographical derivatives obtained from a high-resolution digital elevation model and lithological parameters were employed for the production of landslide susceptibility maps. For the integrated hazard susceptibility assessment, the Mamdani fuzzy algorithm was considered, and the results are discussed in the present study. The results demonstrate that multi-hazard susceptibility assessment maps for urban planning can be obtained by combining a set of expert-based and ensemble learning methods. [ABSTRACT FROM AUTHOR]