KLASIFIKASI GAMBAR TERHADAP IDENTIFIKASI POTENSI SUKABUMI DENGAN METODE DEEP LEARNING LOGISTIC REGRESSION

This research focuses on image classification, namely grouping images that have the same characteristics or the same characteristics. In this study, what is classified is the image with the most classes for any field that has potential with a high percentage to potential with a low percentage, so that by knowing the potential that exists can get the right handling and reference for potential development in the future. The author discusses the methods used and needed in achieving the research objectives, namely combining the Logistic Regression method in Orange Data Mining to increase accuracy and precision in data classification. The use of combined Logistic Regression methods can make the analysis process much more complex. In this stage of testing the classification results, this research uses the Hierarchical Clustering widget to see the classification results of image data in 7,777 photos. From the results above we can see which categories have the most potential, namely Tourism (C3) has the greatest potential, followed by Culinary (C2) and Culture (C4) while Handicrafts (C1) shows the least potential. From all stages of the research that has been done, the conclusion obtained is that the classification of images with 4 predetermined categories using the Logistic Regression method in Orange Data Mining has succeeded in identifying the most influential aspects in Sukabumi through images on Instagram using Orange Data Mining Tools, test data of 7,777 images that have been classified through the Hierarchical Clustering widget, the biggest result is the Tourism category, followed by the culinary category, culture and the least is the craft category.

Keywords : Clasification, Image Data, Orange Data Mining, Logistic Regression, Hierarchical Clustering.