Search

One of the main causes of death in the world is caused by cancer. And one of them is Lung Cancer. According to World Health Organization, in 2014 the death rate caused by lung cancer in Indonesia was 21.8% in men and 9.1% in women with 30,865 cases of death caused by lung cancer each year in men and women. Many studies have been carried out on computational lung cancer. One of them is by implementing machine learning in detecting lung cancer. However, there are obstacles, namely the imbalance in the amount of data between patients and non-patients. So it takes an approach to overcome this imbalance. One of these methods is ADASYN which is then combined with the Random Forest classification algorithm. In this study, we will compare the results of the classification model performance in Random Forest before and after the ADASYN sampling method was used in the training process. The results of this study showed an increase in the performance of the Random Forest classification model with an AUC value of 0.859 after the ADASYN sampling method was carried out with an error rate of 4.9%. [ABSTRACT FROM AUTHOR]

Insurance for paddy farmers or Asuransi Usaha Tani Padi (AUTP) and MSME credit or Kredit Usaha Rakyat (KUR) are government's programs to support farmers in Indonesia. However, many farmers still have not accessed KUR due to failing to pass credit checking against the Financial Information Service System. Integration of AUTP and KUR offers a solution to increase KUR accessibility for farmers and supports thebank to manage credit risk. Some selected districts in Indonesia have been chosen to implement the integration program. Therefore, this study aims to map paddy farmers' profile and understand the potential of the integration program across districts in Indonesia. In addition, we analyzed factors that influence the farmers' willingness topay (WTP) for the integration program premium. The biplot analysis proposed four groups among the researched regions. Group 1 consist of Karawang, Simalungun, and West Lombok, Group 2 consist of Malang, Lamongan, Cirebon, Bandung, Central Lombok, and Langkat, Group 3 consist of Kulon Progo, and Group 4 consist of Sleman, Bogor, Sukoharjo, and Klaten. Group 1 and Group 2 are the most favorable groups for the integration program. The median WTP across the regions is IDR 50,000 with an average of IDR 70,343. Based on the random forest analysis, we concluded nine importantfactors influencing the WTP, including; agricultural income, credit source (financial institutions), age, total land area, loanamount, insurance satisfaction, farming experience, non-agricultural income, and formal education. Among those factors,agricultural income, insurance satisfaction, non-agricultural income, and formal education have a positive correlation withWTP. Meanwhile, the other factors are negatively correlated with WTP. [ABSTRACT FROM AUTHOR]

Floods, natural disasters that occur worldwide, have become more and more frequent in recent decades. Floods are one of the most common disasters in Indonesia, especially when rainfall intensity is significant. Based on the history of flood events in Indonesia over 20 years, flooding dominates all existing disasters by 37%. Flood susceptibility mapping is widely recognized as a necessary step in preventing and managing future flooding. Many flood-prone mapping technologies have been developed and implemented in recent years. The main difficulty associated with the flood-prone mapping process is the multi-variable and non-linear relationship between indices and risk level. In comparison to other, more commonly used multivariate regression or clas-sification approaches, the Random Forest technique has significant advantages. The purpose of this study is to use Random Forest to effectively examine the distribution of flood hazard risk in the Sumedang area. The ten flood conditioning factors were identified from the references including distance to a river, elevation, geology, geomorphology, lithology, land use and land cover (LULC), rainfall, slope, soil type, and topographic wetness index (TWI). The model was assessed using the area under the receiver operating curve (AUC). The AUC value is 0.84 and the value is regarded as being quite good. The Random Forest algorithm produced a probability map with values ranging from 0 to 1 after applying it to the flood conditioning factors. The flood susceptibility map was created from the probability map using the Jenks natural breaks categorization approach. Five probability map classifications are represented: very low (20.94%), low (25.33%), moderate (25.58%), high (18.29%), and extremely high (9.86%). Flooded areas were identified in this analysis as the map classes with high and very high probabilities. So based on this study, the flooded areas from the Random Forest model are 28.15% and the rest 71.85% are non-flooded areas. [ABSTRACT FROM AUTHOR]

During the Covid-19 pandemic, internet usage in Indonesia increased enormously. It was reported that online duration for adults increased by 52% compared to before the pandemic. The easiest way to get the latest information about the pandemic is to access online news portals. However, people often skim the unique and bombastic headline, then click on it leading to the whole story. This occurrence is called clickbait. Clickbait is a term that refers to website content in the form of news, advertisement, or services intended to gain attention and encourage visitors to click on links to certain web pages. This study aims to automatically classify clickbait news on online portals using a statistic model, random forest, that contains a collection of decision trees. Because of the unbalanced data, the researchers employ SMOTE to balance the amount of data in the minority class by generating synthesis data. The total of the news is 2,028 of its headline and content about covid. Preprocessing data includes case folding, removing number and punctuation, removing stop words, and lemmatization. The transformation of text into numeric representation using TF-IDF was done separately between headline and content. The best weight for the TF-IDF's headline and content applied before concatenating process is 70%:30%. Based on the analysis results with a ratio of training data and testing data of 80%:20%, the accuracy value is 80%, precision is 99%, and recall is 63%. [ABSTRACT FROM AUTHOR]

This study presents an experimentation result to observe the impact of spatial scales and imbalanced data treatment on the landslide susceptibility mapping using machine learning algorithm namely Random Forest (RF). Three districts in the Central Java Province, Indonesia have been used as the case study with regards to the data availability. The presence of data scarcity of landslide occurrence causes data imbalance at various pixel sizes. To treat imbalanced data, Synthetic Minority Oversampling Technique (SMOTE) resampling technique was employed. The RF algorithm was implemented in two modes namely trained using raw imbalanced dataset and trained using SMOTE-balanced dataset on all designated pixel sizes. Five frequently used metrics were chosen to test the model skills. The results show that balancing the data with SMOTE has notably improved the model performances. The performance of models trained using raw data have no trend of improvement with spatial scale density. On the other hand, the model performance trained using SMOTE-balanced data reveal an enhancement of mapping accuracy along with the increase of spatial resolution. Hence, it is prudent to treat imbalanced data using a suitable method like SMOTE to produce a reliable machine learning-based landslide susceptibility mapping that is valuable for predictive disaster mitigation planning and management. [ABSTRACT FROM AUTHOR]