Flight Satisfaction

Pytorch model to predict customer flight satisfaction based on survey results.

Data

Data was pulled from Kaggle. Code expects test.csv and train.csv to be placed in the "data" directory.

Environment

Create a Python venv and pip install from "requirements.txt".

Data Preprocessing

Before running the model, data preprocessing is done. The following happens:

• Missing values are filled in with 0.
  • These missing values only appear in the "Arrival Delay in Minutes" column and were assumed to be 0 minutes.
• The columns 'Customer Type', 'Type of Travel', and 'Class' are turned into multiple columns with one-hot encoding.
• The 'Gender' column is turned into Gender_Female, a boolean column representing if the customer is female or not.
• The 'satisfaction' column has its values turned into 1 for "satisfied" and 0 for "neutral or dissatisfied".