World Energy

---

Summary of Mean Squared Error and R-Square

Overall, the historical predictions are stronger than the future predictions. The machine learning models Multiple LinearRegression and Ridge impacted the oil production and consumption R-Square results and historical MSE differently than the rest of the models. Multiple LinearRegression and Ridge are inferior to Lasso and Elastic Net in Oil predictions however they are all fairly equal within the natural gas predictions. The R-Square results for the historical data that passed through the natural gas prediction models are all relatively close. Among the natural gas predictions, based purely on historical data, the Ridge machine learning model has the greatest fit.

Moving averages were used to generate future features and did not fare well across all models running future predictions. The features show more dimension on the data and running predictions thereafter.