Flight prices can feel arbitrary—until the pattern comes into view. I trained a model on 9 million flight records to estimate a round-trip price from the details a traveler can actually name: route, airline, and ticket count.
Choose an origin, destination, airline, and ticket count; the model returns a round-trip estimate. The linked notebook and repository make the modeling and statistical work behind that simple question inspectable—because a prediction is more useful when you can see how it was made.
Flask, Heroku, Dash Core Components, Scikit-Learn, Category_Encoders, PDP (Partial Dependence Plots), Plotly, Matplotlib, Seaborn, Shap (Shapley), eli5 (Permutation Importances)
Solo Project
Less than Two Weeks
