DATA MINING USING CRISP-DM PROCESS FRAMEWORK ON OFFICIAL STATISTICS: A CASE STUDY OF EAST JAVA PROVINCE

Data mining on official statistics becomes a study interest, as it offers an opportunity to reveal hidden patterns within the data. This study investigates the data mining process's appropriateness using the CRISP-DM method to a secondary-quantitative data analysis and to investigate hidden information revealed from data mining on official statistics. Data is collected from the East Java BPS website, and the unit of analysis is regency/municipality. Five macro development indicators (Human Development Index, Gross Regional Domestic Products, poverty rate, Gini Ratio, open unemployment rate) are selected as analysis variables. Workflows of data analysis are designed using Knime software.  This study shows the usefulness of the CRISP-DM method for secondary research because it specifies standardized stages for analyzing secondary data and improves the secondary analysis rigor. Furthermore, the clustering technique classifies regencies/municipalities into three clusters. One of the clusters has desirable indicator levels: high Human Development Index - high Gross Regional Domestic Products - low poverty rate, together with undesirable ones: high Gini Ratio - high open unemployment rate. This result indicates that a regency/municipality might not achieve an ideal condition of the five macro development indicators. Some indicators such as the open unemployment rate might be an inevitable impact. This research adds to the literature on development economics studies, particularly on the application of data mining, the CRISP-DM method, and Knime software to official statistics.