Analysis of restaurant salesthrough machine learning

The restaurant business is a volatile market riddled with preconceived notions making it a difficult market to do scheduling and inventory management in. Using machine learning algorithms we can ease the difficulty of these tasks through better understanding of the market and the possibility to more accurately estimate sales. Using a random forest regressor we are able to predict sales to 0.9135 R2 and approximately 6983 RMSE. Making random forest regression a viable algorithm to predict sales. Through more features such as weather it is believed that the algorithm can be improved further. Through analysing the data and its trends it is concluded that sales go up significantly during the weekends. Sales go down during January and August-November and go up during February-July with the highest spending in December. During the summer months the spending have less variation due to summer vacations allowing people to spend more time and money at restaurants during the weekdays