Potable water is water that is fit or suitable for drinking. The benefits of drinking potable water include increasing energy, reducing fatigue, supporting weight loss, removing toxins from the body, and assisting in the digestion of food and prevention of constipation. On the other hand, drinking contaminated water can cause people to have diarrhea, stomach cramps, vomiting, and pneumonia. If you drink water with contaminants like arsenic, you can increase your chance of developing cancer. Although water treatment methods like boiling water for at least an hour, filtering water through ceramic filters, and filtering water through simple carbon-activated filters can remove many contaminants from tainted water, they cannot take out all the contaminants from the water. However, I plan to use the continuous features of potable water such as pH, capacity of water to precipitate soap in mg/L, total dissolved solids in ppm, amount of Chloramines in ppm, amount of Sulfates dissolved in mg/L, conductivity, amount of organic carbon in ppm, amount of Trihalomethanes in μg/L, turbidity in making a model to predict the discrete target of potability. This model will have a feature layer as an input layer, two dense layers that used 128 nodes and rectified linear unit activation, a dropout layer, and a dense output layer with one node. To make this model have a higher chance of accurately predicting potability, I plan to experiment with different combinations of these numerical features and crossed feature columns in the model. This model will help enjoy the benefits of potable water and avoid the negative effects of contaminated water. If I can figure out what combination of features helps increase the model’s accuracy, I could find out the features necessary to differentiate between contaminated and potable water. In addition, this discovery could lead to a way to perfectly convert contaminated water into potable water. If this model is not able to achieve a high enough accuracy value, I would recommend observing the random forest classifier and using the standard scaler, more dense hidden layers, more features, less inaccurate potable water observations to improve this model.