Predcition-of-Airline-Passengers'-Satisfaction

This project aims to compare different classification models (Logistic regression, Decision tree, and SVM) to predict airline customers' satisfaction.

Resources Used

• Python Version: 3.7
• Packages: pandas, numpy, sklearn, matplotlib, seaborn, selenium.
• ChromeDriver 95.0.4638.10 download.

Data used

• Airlines Customer satisfaction download.
• This data shows whether a customer is satisfied with the airlines or not based on different 9 attributes:
  1. Gender
  2. Customer Type
  3. Age
  4. Type of Travel
  5. Class
  6. Flight Distance
  7. Seat comfort
  8. Departure/Arrival time convenient
  9. Food and drinks

Data preparation

• Removed missing values.
• Sampled data for analysis and visualization.
• Visualised correlation between attributes:
• SVM is a slow algorithm, thus I used a sample of 5000 rows from the original data as opposed to 100.000 samples used for the logistic regression and Decison tree model.

Classifcation models:

Logistic regression:

• Mean Absolute Error: 0.161
• Accuracy: 0.84

Decision tree:

• Mean Absolute Error: 0.062
• Accuracy: 0.94

Support vector machine:

• Mean Absolute Error: 0.092
• Accuracy: 0.91