The Classification of Poor Households in Jombang With Random Forest Classification And Regression Trees (RF-CART) Approach as the Solution In Achieving the 2015 Indonesian MDGs Targets

Bambang Widjanarko Otok ^[3] | Dian Seftiana

Abstract: Poverty is the main cause from some social, politic, even economy problems, especially in the developing countries. Poverty alleviation is the main key done to achieve the targets in 2015 Millennium Development Goals (MDGs). The government assistance program as an effort in poverty alleviation has not yet been able to decrease the amount of poor people in Jombang regency, in fact that Jombang regency has positive economy grow rate from year to year. So, to make the government assistance become effective and efficient, it is done the classification based on the household assistance package. It is expected that by using CART approach and combining with Random Forest method, the classification of poor housseholds in Jombang regency can be more accurate. The classification result shows that by using most important variable in determining, the desired assistance package is type of cooking fuel for Chronically Poor Household and montlhy income for Poor Household. The total accuracy rate (1-APER) resulted from the used of CART method is 0, 4313 for Chronically Poor Household and 0, 4338 for Poor Household. While, the classification remedial by using RF-CART method results importance variable that is the monthly income with 0, 9950 total accuracy rate for Chronically Poor Household and 0, 9833 for Poor Household. While, RF-CART is the better method in classifying poor households in Jombang regency because it can increase the total accuracy rate of 0, 5637 for Chronically Poor Household and 0, 5495 for Poor Household.

Keywords: CART, MDGs, Random Forest, Total Accuracy Rate

Edition: Volume 3 Issue 8, August 2014,

Pages: 1497 - 1503