Conclusions

Overall, it's difficult to make any useful conclusions from the results of this project. 

One could say that all models are "better" at predicting higher temperatures than lower. In all of the classification models, the prediction frequency and accuracy for "above" and "extremely above" average for global average temperature was higher than the other class labels (i.e. "below" and "extremely below").  And then the regression models had particularly higher accuracy scores for maximum and average temperature. 

However, this was likely due to the distribution and discretization of the utilized data more so than the quality of model performance.