Prediction of Clean Water Needs in Surabaya Using the Time - Series Neural Network Method

The need for clean water every year generally continues to increase, but on the other hand the availability of clean water is increasingly limited due to the narrow catchment area. Java Island itself is classified as a water-critical island, where each resident on the island of Java can only meet their water needs of 1750 cubic meters. This value is below the standard which should be 2000 cubic meters per capita per year. When viewed from the composition and burden of needs that must be provided, the island of Java, which is only 7% of the total land area in Indonesia, but is inhabited by 65% of Indonesia's population, could experience water shortages. This is because water needs on the island of Java require at least 45% - 55% of water resources, while the potential for water resources on the island of Java is currently only 4.5% of the total potential of natural resources. If it continues, it is predicted that there will be water scarcity in 2040. Surabaya as the research target is the second largest city in Indonesia after Jakarta. Based on the National Social and Economic Survey of 2011, the consumption of water that comes directly from Regional Water Supply Company taps is 70% followed by the use of groundwater in protected wells, which is affected by rainfall. Based on this problem, the utilization of the Machine Learning Time-Series Neural Network method is used and researched to predict future water demand in Surabaya. The prediction results show that in 2040, regional water company must meet the water needs of 625,253,727 cubic meters for 1,457,326 customers, or 429,000 liters of water per customer. The results also show that when compared with the linear regression method, the Time-Series Neural Network method has a better MSE.