This project conducted for National Data Science Tournament 2021 in Indonesia and we've managed to win the tournament as the third winner. Our team created a dashboard with Tableau Public based on the modeling of the number of poverty in Jawa Barat using multiple linear regression with dummy variables for each districts/cities. This particular dummy variables created to shown the marginal effect of each districts/cities in Jawa Barat to the number of poverty in its region.
Our team called AFIB which consider to be the abbreviations of our member that consists of:
Data we've used came from Badan Pusat Statistik (BPS) that have been compiled into the Open Data Jabar.
From the modeling results we can extract some insights:
- There are 5 districts that have the highest marginal effect to the number of poverty, which is Kabupaten Bogor, Kabupaten Bandung, Kabupaten Cirebon, Kabupaten Indramayu, and Kabupaten Karawang.
| Districts | Estimated Coefficient |
|---|---|
| Kabupaten Bogor | 401,438418 |
| Kabupaten Bandung | 244,401710 |
| Kabupaten Cirebon | 223,586339 |
| Kabupaten Indramayu | 212,567116 |
| Kabupaten Karawang | 207,245554 |
- Besides of the dummy variables, there are three predictors that has significant effect on the number of poverty in Jawa Barat. There are pengeluaran_perkapita, apk_perguruan_tinggi, and tingkat_pengangguran_terbuka.
| Predictors | Estimated Coefficient |
|---|---|
| pengeluaran_perkapita | -40,9845 |
| apk_perguruan_tinggi | 8,3953 |
| tingkat_pengangguran_terbuka | 18,3210 |
We've also made the coefficient and partial dependence plot in our google colab notebook to gain more insight from the estimated model.