Ananto, Mochamad Ihsan (2019) Klasifikasi Kategori Pengaduan Masyarakat Melalui Kanal LAPOR! menggunakan Artificial Neural Network (ANN). Other thesis, Institut teknologi Sepuluh Nopember.
Preview |
Text
06211540000123-Undergraduate_Theses.pdf Download (1MB) | Preview |
Abstract
LAPOR! merupakan sarana aspirasi dan pengaduan masyarakat terkait kinerja pemerintah berbasis media sosial. Oleh karena laporan pengaduan masyarakat yang masuk tersebut berbentuk teks, maka dapat diselesaikan dengan cara text mining. Sehingga dilakukan analisis klasi-fikasi teks menggunakan Artificial Neural Network serta SMOTE untuk mengatasi data imbalance dan Chi-Square untuk proses seleksi variabel. Data yang digunakan adalah data historis aduan masyarakat melalui kanal LAPOR! tahun 2015. Melalui proses seleksi variabel, didapatkan sejumlah 428 term atau kata yang memberikan pengaruh terhadap ka-tegori aduan masyarakat. Ketepatan klasifikasi yang dihasilkan melalui metode Artificial Neural Network dengan feature selection dan 3 nodes hidden layer adalah precision 0,794, sensitivity 0,818 dan F1-Score 0,800. Selain itu didapatkan topik permasalahan yang patut mendapatkan perhatian lebih pada kategori aduan energi, pangan dan maritim adalah kata raskin. Untuk kategori infrastruktur dan transportasi adalah aduan mengenai jalan. Pada kategori kesehatan, yakni kata BPJS. Untuk kate-gori pendidikan adalah kata terima yakni berkaitan dengan pembagian KIP. Lalu untuk reformasi birokrasi adalah KTP. Sedangkan untuk pari-wisata dan lingkungan hidup adalah kata imigrasi.
================================================================================================================================
LAPOR! is a media of aspirations and public complaints related to the performance of government based on social media. Because of public complaints received in the form of text, it can be solved by text mining. Therefore, the text classification analysis is used with application of Artificial Neural Network and SMOTE to overcome imbalance data and Chi-Square for the variable selection process. The data used is the historical data of public complaints through LAPOR! canal in 2015. Through variable selection process, obtained 428 terms or words which give effect to the category of public complaints. The performance measure of classification produced through the Artificial Neural Network method with a feature selection and 3 hidden layer nodes precision 0,794, sensitivity 0,818 dan F1-Score 0,800. In addition, the topic of the issue that deserves more attention in the category of energy, food and maritime complaints is the raskin word. For the category of infrastructure and transportation, there is a complaint regarding the road. In the health category, BPJS. For the education category, the word accept is related to the distribution of KIP. Then for bureaucratic reform is KTP. While for tourism and the environment is the word immigration.
| Item Type: | Thesis (Other) |
|---|---|
| Additional Information: | RSSt 519.53 Ana k-1 2019 |
| Uncontrolled Keywords: | Artificial Neural Network, LAPOR!, SMOTE, Text Mining, Word Cloud |
| Subjects: | Q Science > QA Mathematics > QA278.55 Cluster analysis Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) |
| Divisions: | Faculty of Mathematics and Science > Statistics > 49201-(S1) Undergraduate Thesis |
| Depositing User: | Ananto Mochamad Ihsan |
| Date Deposited: | 29 May 2023 02:57 |
| Last Modified: | 29 May 2023 02:57 |
| URI: | http://repository.its.ac.id/id/eprint/63929 |
Actions (login required)
 |
View Item |