Claim Missing Document
Check
Articles

Found 2 Documents
Search
Journal : Journal of Applied Business and Technology

Comparison of Feature Selection with Information Gain Method in Decision Tree, Regression Logistic and Random Forest Algorithms Sholeh, Muhammad; Lestari, Uning; Andayati, Dina
Journal of Applied Business and Technology Vol. 5 No. 3 (2024): Journal of Applied Business and Technology
Publisher : Institut Bisnis dan Teknologi Pelita Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35145/jabt.v5i3.155

Abstract

One of the approaches that can be done is to perform feature selection. Feature selection is done by identifying the most informative features and not using features that do not directly contribute to the target feature. The purpose of feature selection is to increase the accuracy of the model. The research was conducted by comparing the performance of the model by comparing the accuracy results of the model without any feature selection with the model that has done feature selection. The process is done by comparing the accuracy results with decision tree, random forest and SVM algorithms. In the research method of feature selection on science data, the steps include understanding the domain and dataset, exploratory analysis, data cleaning, measuring feature relevance with criteria such as Information Gain, and feature ranking. The results are evaluated and validated using model performance metrics before and after feature selection. This process ensures selection of relevant features, improving accuracy. The research process used the Lung Cancer Prediction datasheet which consists of 306 rows and 16 attributes. The results show that feature selection can improve the performance of the classification model by reducing features that do not contribute to the target. Comparison results using decision tree, Regression Logistic and random forest classification model algorithms and feature selection resulted in a high accuracy value of 0.968 in the Regression Logistic algorithm with a feature selection of 5.
Application of K-Means Algorithm in Clustering Model for Learning Management System Usage Evaluation Sholeh, Muhammad; Suraya; Dina Andayati
Journal of Applied Business and Technology Vol. 4 No. 3 (2023): Journal of Applied Business and Technology
Publisher : Institut Bisnis dan Teknologi Pelita Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35145/jabt.v4i3.130

Abstract

The use of a learning management system (LMS) is one of the media that can be used to disseminate lecturer materials to students. Materials that can be uploaded on the LMS can be in the form of lecture materials in the form of files, videos, or questions. The effectiveness of LMS can be evaluated by looking at activities in using LMS. The effectiveness of using LMS can be seen from the log. Log results from LMS can be evaluated in various ways and one way is to use data mining clustering models. The clustering model can be used to create student groupings and the clustering results can be labeled in the form of categories, such as very good, good, and bad categories. This labeling depends on the clustering results that will be processed in the modeling. The research method uses CRISP DM which consists of business understanding, data understanding, data preparation, modeling, evaluation, and deployment. The beginning of the research process is carried out by taking log data in the Moodle LMS. The clustering model in this research will use the K-Means algorithm and the evaluation of clustering results will be evaluated for performance using the Davies-Bouldin method. Implementation of data mining processing using Rapid Miner application. The datasheet used is a datasheet taken from the LMS log of the Computer Programming course in the Mechanical Engineering study program - AKPRIND Institute of Science & Technology Yogyakarta odd semester of the 2021/2022 and 2022/2023 academic years. The results of the study resulted in the best clustering based on the Davies Bouldin method of 2. The clustering results, cluster 0 consists of 28 data named the category of frequent access to LMS and cluster 1 consists of 54 with the category of not frequent access to LMS.