ANALISA DATA MINING DALAM MEMPREDIKSI MASYARAKAT KURANG MAMPU MENGGUNAKAN METODE K-NEAREST NEIGHBOR

Abstract

Poverty is one of the fundamental issues that is center of attention of the government in a country. One important aspect to support the poverty reduction strategi is the availability of accurate and targeted poverty data. One of the main problems that often hinders the success of these government programs is the availability of appropriate data on the targeting of the poor. This study aims to design an application than can predict the poor using the K-Nearest Neighbor Algorithm with the five main indicators being the type of work, number of dependents, age income and condition of the household head of the family. This prediction provides data on poor families that are suitable for receiving various assistance from the government. The data used for predictions are sample data from Pegasing District. In this study, the K-NN Algorithm was analyzed which was developed based on the web. The working principle of K-Nearest Neighbor is to find the shortest distance between the evaluated data and training data. The results of the evaluation using the confusion matrix obtained the resulting accuracy for 216 training data with 93 testing data with a ratio of 70:30 and five attributes used produced an accuracy of 86,02%, Recall 61,90%, Precision 72,22%, and F1-Score 66,04%.