Predicting Airlines customer satisfaction

Project for the Quantitative Analysis for Managerial Decisions course.

Project description

The project focuses on the decision-making workflow behind the use of mathematical models to make informed answers about a problem at hand. In the case of our project, we faced a supervised binary classification problem of predicting the overall customer satisfaction based on several features - objective measures of satisfaction provided by customers of the anonymized Invistico Airlines.

We attempt to build a classification model that is both:

• Accurate in it's predictions
• Explainable - to help guide the company actions to maximize the number of satisfied customers

TOC

1. Data preprocessing
   • EDA
   • Missing data
   • Feature Engineering
   • Feature Encoding
2. Model fitting and validation
   • Adaline
   • Random Forest
   • Logistic Regression
   • Neural Network
3. Model performance comparison
4. Summary

How to?

• Please refer to the file AirlinesCustomerSatisfaction.pdf for the full generated report.
• Please refer to the file AirlinesCustomerSatisfaction.Rmd for the detailed code used to generate it

Authors:

• Joanna Krężel
• Anna Matysek
• Adam Szczerba

• Piotr Mikler