Access to fresh water has become a headache for many nations around the world today due to water scarcity. Since this aspect is increasing, this paper introduces a system to estimate the households’ water needs in a day taking into consideration, the size of the family, zone, temperature, season, working status, location, and religion. By applying machine learning, the system determines the balance of water distribution between the families by these parameters. The predicted values determine the water requirements for each household hence the distribution of water in the households on a particular day. This includes the use of an Arduino microcontroller, water flow meter, and solenoid valves that make the water supply system to be an automatic system. As a result of this, the system helps to control the wastage of water and also ensures that the amount of water required in each home is assessed accurately.