Village Classification based on Geographic Difficulties using Backpropagation Neural Network Algorithm (Case Study: Village Potential of Sumenep Regency)

Indonesia has a variety of geographical conditions, from coastal islands, hills, and mountains. For an archipelago consists of 17, 504 islands, the diversity of the archipelago affects the equitable development process. Until recently we have seen a growth center in the region with a flat topography. Developments in hilly, mountainous, and archipelagic regions can be expensive. Sumenep Regency is one of the districts still lagging in East Java Province, and its characteristics are more complex than other districts in East Java. This research was conducted to obtain information about attributes that have an important role as a reference for village development priorities by classifying the level of geographical difficulties based on indicators used as attributes. The initial test with 28 attributes and after features selection using Chi-square and Cramer's V correlation became 25 attributes the accuracy reaches 87, 62% compared to the initial attributes with the classification results reaching 83, 07 %.