Data Analytics of Human Development Index(HDI) with Features Descriptive and Predictive Mining

The value of the Human Development Index (HDI) in Indonesia is increasing every year. Indonesia has many provinces and districts/cities, it makes the government need more time to analyze data. This research purpose a new method to analyze the data of HDI with a descriptive and predictive mining method. There are two main results of this research. First, a segmentation of HDI data into four segments, there are low, medium, high, and very high. Second, a prediction of HDI data. Before analyzing data, the system does data preprocessing to repair the missing data (cleaning) and normalization (transformation) to convert data into a smaller range(from 0 to 1). To get a segmentation result use the descriptive mining method, in this method, there are two steps, the first system does grouping and labeling data based on the value of HDI indicators(life expectancy, expected years of schooling, mean years of schooling and income per capita) use Hierarchical Clustering Centroid Linkage Method. Second, the system does the interpretation process based on the distance between centroid every cluster and ground(0,0). To get a prediction result use the predictive mining method, this process uses a Weighted Moving Average(WMA) with the last three years of HDI data. The result of this research, the variance accuracy value of the descriptive mining method is 0,203, and the Mean Absolute Percentage Error(MAPE) value of the predictive mining method is 0,27%.