Decision Tree Classifier for Market Share Prediction

Inspiration

Spirit Airlines has faced significant financial challenges, including low profits, stock price drops of 83%, and failed mergers. With 5 out of the last 6 quarters reporting losses and net income losses of $192,927 in 2024 Q2, the airline needs a strategic approach to identify routes where it can compete effectively against larger airlines. This project is inspired by the need to optimize Spirit’s market share on profitable routes, leveraging machine learning techniques to pinpoint potential opportunities.

What it does

This project uses a Decision Tree Classifier to predict the market share of the largest airline carrier on each route, categorizing it as High, Medium, or Low. By identifying routes where the market share of the largest carrier is low or medium, the model reveals strategic opportunities for Spirit Airlines to enter the market and potentially increase its market share. The model focuses on optimizing Spirit's routes by highlighting areas with the greatest potential for revenue growth.

How we built it

We began with an exploratory data analysis using tools such as Numpy, Pandas, Seaborn, and Matplotlib to explore the dataset’s structure. During this phase, we identified correlations between the variables but realized that the data was non-linear. To confirm this, we applied Principal Component Analysis (PCA), which validated the non-linearity in the dataset.

To model this non-linear dataset, we implemented a Decision Tree Classifier to categorize the 'large_ms' variable into three distinct groups: 'Low', 'Medium', and 'High'. The model utilizes Scikit-learn's capabilities to perform the classification task and evaluate its performance using cross-validation.

Challenges we ran into

We had issues finding a significant research question. Additionally, we had problems executing the PCA.

Accomplishments that we're proud of

We successfully built a model that can predict market share with a high degree of accuracy, providing actionable insights into which routes present strategic opportunities for smaller airlines like Spirit. We identified routes where low or medium market share presents a viable opportunity for Spirit Airlines to compete against larger carriers.

What we learned

Machine Learning for Competitive Strategy: How decision tree models can be used to make strategic decisions in competitive industries like air travel.
Optimizing Routes for Profitability: The importance of route optimization in helping smaller airlines identify competitive opportunities.
Feature Importance: How key features such as fare, passenger count, and distance impact the market share dynamics of airline routes.