Stochastic Prediction of Small-Town Lottery Numbers using User-Based Collaborative Filtering

There has been a great effort to model and predict the randomness of lottery games. Modeling the randomness of the winning numbers is extremely difficult as there is a small to no chance of winning the lottery. In this research, we propose a method to predict the Small-Town Lottery numbers using the User-Based Collaborative Filtering approach. We gathered 1,599 STL winning combinations from November 17, 2017, up to May 21, 2019, in the Philippine Charity Sweepstakes results website. Each winning combination is represented as users, while the winning numbers are accounted as items in the model. The objective is to predict the next winning combination using the last draw result through ratings. The ratings were based on the draw schedule of the winning combinations. Based on the experiments conducted, the model was able to achieve a 73.55% Accuracy Score in the 10-fold cross-validation while managing a 0.66 Root Mean Square Error, and 0.40 Mean Absolute Error in the split-validation. This indicates that the method could significantly increase the odds of winning while extensively capturing the randomness of the game.