British Airways Project

In progress....

This project was part of a virtual experience program at British Airways.

Task 1:

Scrape and analyse customer review data to uncover findings for British Airways.

Task 2:

Build a predictive model to understand factors that influence buying behavior.

Task 1

First data was scrapped from Skytrax website. This was my first attempt at scrapping data and I was sucessful in scraping most of the data. However I was struggling with scraping the star ratings that users gave. To help, I used this notebook.

The next stages inovled cleaning the data and checking for missing values. Futher details on ths can be seen in the notebook.

Exploratory Data Analysis

How has overall customer ratings for British Airways evolved over time?

Observation

After decomposing the time series additively, we see a clear and consistent decreasing trend in the overall rating from customers over the period of 2015 to 2023.
The residuals are centered around 0, which supports the use of an additive model.
The noise observed from mid-2020 to mid-2021 in the residuals may be attributed to the linear interpolation used to fill in the missing values during that period.

Distribution of review ratings for British Airways

Observation

Looking at the distribution for overall rating, we see that most people give a rating of 1. Other than that, there does not seem to be any difference between the other ratings.
We see that most guests are not satisfied with the WiFi service.
The staff service receives high ratings, with most people giving 4 and 5 stars.
There seems to be a uniform distribution of seating comfort.
Most people are writing between 500 to 1000 words per review.

Length of reviews per travel class

Observation

We see, there does seem to be an unbalanced dataset. There are alot more people that traveled Economy and Business in this dataset. The distributions follow the same shape between different travel classes.

Overall rating given per travel class