Deep Learning untuk Penilaian Kelayakan Penerimaan Bantuan Sosial Berdasarkan Foto Rumah

Nevin, Muhammad (2022) Deep Learning untuk Penilaian Kelayakan Penerimaan Bantuan Sosial Berdasarkan Foto Rumah. Project Report. [s.n.], [s.l.].

Text Buku KP_05111940000079_Muhammad Nevin.pdf - Accepted Version Restricted to Repository staff only Download (1MB)

Abstract

Dalam rangka perbaikan kualitas data, Kementerian Sosial melakukan verifikasi dan validasi data penerima bantuan sosial melalui aplikasi aplikasi Social Affair Geographic Information System (SAGIS). Sistem ini merekam data berupa kuesioner tentang kondisi ekonomi dan data berupa foto rumah keluarga penerima manfaat (KPM). Foto rumah tersebut secara visual dapat memberikan kesan apakah KPM termasuk layak atau tidak menerima bansos. Oleh karena itu diperlukan suatu sistem yang dapat memberikan rekomendasi penilaian kelayakan penerimaan bantuan sosial secara otomatis berbasis foto rumah. Image Classification adalah metode mengelompokkan objek berdasarkan class (kelas) tertentu berdasarkan hasil ekstraksi informasi citra. Dua di antara sekian banyak model deep learning untuk klasifikasi gambar adalah CNN dan VGG16. Tujuan dari KP ini adalah menghasilkan model klasifikasi gambar yang dapat membantu Kementerian Sosial meningkatkan efisiensi kinerja untuk menilai kelayakan penerima bantuan sosial. Proyek ini menggunakan data pada wilayah Jawa Timur dengan total 3847 data layak dan 3849 data tidak layak. Untuk meningkatkan variasi, dilakukan augmentasi data pada data train sehingga menambah jumlah data train menjadi 3 kali lipat. Berdasarkan percobaan, didapatkan akurasi model CNN sebesar 58,44% dan VGG16 sebesar 64,75%. Proyek ini berhasil menentukan threshold optimum dalam penentuan label prediksi dengan < 0,05 untuk label layak, > 0,99 untuk label tidak layak, dan sisanya untuk label ragu-ragu sehingga dapat meningkatkan efisiensi kinerja sebesar 53,6%.
===========================================================================================================================
In order to improve data quality, the Ministry of Social Affairs verifies and validates social assistance beneficiary data through the Social Affair Geographic Information System (SAGIS) application. This system records data in the form of questionnaires about economic conditions and data in the form of photos of beneficiary families' houses (KPM). The photo of the house can visually give an impression whether the KPM is eligible or not to receive social assistance. Therefore we need a system that can provide recommendations for evaluating the eligibility of receiving social assistance automatically based on house photos. Image Classification is a method of grouping objects based on a particular class based on the results of image information extraction. Two of the many deep learning models for image classification are CNN and VGG16. The purpose of this KP is to produce an image classification model that can help the Ministry of Social Affairs improve performance efficiency to assess the eligibility of social assistance recipients. This project uses data from the East Java region with a total of 3,847 feasible data and 3,849 non-feasible data. To increase the variety, data augmentation was carried out on the train data so as to increase the amount of train data to 3 times. Based on the experiment, the accuracy of the CNN model was 58.44% and VGG16 was 64.75%. This project succeeded in determining the optimum threshold in determining predictive labels with <0.05 for feasible labels, > 0.99 for unfeasible labels, and the rest for doubtful labels so as to increase performance efficiency by 53.6%.

Item Type:	Monograph (Project Report)
Uncontrolled Keywords:	SAGIS, klasifikasi gambar, VGG16
Subjects:	Q Science > QA Mathematics > QA76.6 Computer programming. Q Science > QA Mathematics > QA76.9.D33 Data compression (Computer science) Q Science > QA Mathematics > QA76.9.D343 Data mining. Querying (Computer science) Q Science > QA Mathematics > QA76.F56 Data structures (Computer science) T Technology > T Technology (General) > T57.5 Data Processing
Divisions:	Faculty of Information Technology > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Muhammad Nevin
Date Deposited:	02 Aug 2023 07:54
Last Modified:	02 Aug 2023 07:54
URI:	http://repository.its.ac.id/id/eprint/102078

Actions (login required)

View Item