Conclusions

In summary, Naive Bayes is one of the most powerful tools in the world of supervised machine learning. One big advantage is that they're easy to understand, making them popular in many industries. In this context, Naïve Bayes was employed to predict whether the delay due to flights is short or long based on the origin state, destination state, origin temperature, and destination temperature. The data was cleaned and pre-processed so it could be modeled using Naïve Bayes.  However the result is not super satisfactory, and the performance is almost as same as decision trees, making it not the optimal algorithm for this dataset.  The accuracy is 55% and the F1 score of the model is 0.61 which is not so good. 

The one thing that both decision trees and the Naïve Bayes classification algorithm show is that temperature is not enough to predict the weather delay which implies maybe adding additional features can improve the performance of the models. Or maybe the relationship between the input feature and target variable is so complex that other classification algorithms that can capture complex relationships may perform better. Therefore, exploration of other models or usage of boosting may yield in better performance.