How do the amenities offered (cabin, seat, lounge and over time) impact the flight’s overall rating?

by macaRoni

Summary

We used a Skytrax Reviews dataset with individual data frames for Airline, Seat and Lounge ratings. Each dataset has 20 variables, there are approx. 40000 observations for Airline, approx. 2000 for Seat and Lounge.

• Columns consist of Airline Information, Author Information, Overall Rating, Contributors to the overall rating

• Each observation is a passenger that filled out a survey based on their experience of the airline, lounge or seat.

Question 1: How does the overall rating for airlines change depending on the time of the year?

Parameters:

• Time ranges from 2014 - 2015 (lack of data points for other years)

• Overall ratings averaged by date to avoid multiple data points for the same day

• Top five airlines were chosen with highest overall ratings

Airlines: Air Astana, Asiana Airlines, Garuda Indonesia, Bangkok Airways, Indigo Airlines

Main finding: no significant pattern in between years or at certain times of the year as airlines offer the same service throughout each year.

Decrease in overall ratings during the winter months (less survey submissions and therefore negative surveys having more weight).

Limitations:

• Variability of ratings – uncertainty in geom_smooth function

• Limited time period (2 years)

Question 2: To what extent do passengers in First Class(FC) give better ratings than passengers in other cabin types?

We looked at different cabin classes in a flight, specifically the FC cabin since we assume it offers the greatest number of services which others may not offer. The passengers’ experience was measured in terms of the overall rating.

Hypothesis: FC gives the highest overall (flight) rating since its generally known for offering the plushest services.

Result: Business Class, followed by FC

Suggested claim: Fewer people were seeking first class amenities.

Further analysis affirms our claim. This limitation is important to note because out of all the travelers, least number of them chose to fly in the FC cabin.

Using text analysis, we only looked at facilities mentioned commonly in the comments submitted assuming, if the passenger decided to mention it in the feedback, then it may be a stronger factor contributing to the overall rating submitted than the other facilities offered.

To conclude a connection between amenities and overall rating, we looked at summary statistics and visualized variability using density plots which confirmed the values obtained. The overall rating (1-10) and amenities (1-5) have a similar rating proportionately (between 65-70%). Few passengers rated food 1/5 which explains why the average food rating is not in the similar range as others.

Question 3: Improvements in what amenities correlate to improvement in the overal rating the most.

Airline’s improvement is defined as the sign of the slope of the overall rating vs date fit.

The same definition of improvement was used to find the improvement of cabin staff, food and beverage, seat comfort inflight entertainment and value for money rating. We omitted wifi rating due to lack of data points.

The improvement was calculated for every distinct airline reviewer pair. This yields an answer to the question has the airline improved over time in the eyes of a particular reviewer? (yes, no, it got worse)

Our findings

• Cabin staff rating correlates the strongest with the improvement in overall rating

• Inflight entertainment rating correlates the least

Question 4: Do positive reviews on an airline and positive reviews on lounge and seat have a correlation?

Hypothesis: seat and lounge rating to have a positive correlation to the airline rating because both are provided by the airline so this would affect the authors rating of it.

First, we produced graphs to show the percentage difference between the ratings of lounge and airline, and seat and airline. In general, the percentage differences were quite low despite there being a few exceptions as can be seen on the graphs.

We found the correlation coefficient to try and answer the question and it showed that both values are closer to 1 than -1 so both the lounge and seat ratings have a positive correlation to the airline rating which proves our hypothesis to be correct. This suggests that the seat and lounge will have an effect on the authors rating of the airline.

Conclusions:

1. Time doesn’t affect the overall ratings as airlines give same service on all flights.

2. There exists a connection between the cabin class assumed to have the most amenities (FC) and overall ratings.

3. Cabin staff rating correlates the strongest and, in-flight entertainment the least, with the improvement in overall rating.

4. The Seats and lounges will affect the airlines overall rating.

Presentation

Our presentation can be found here.

Pre-Recorded Presentation

Our pre-recorded presentation can be found here.

Password: MacaR0n!

Data

Nguyen, Q 2015, Skytrax User Reviews Dataset (August 2nd, 2015), electronic dataset, quankiquanki, viewed 01 December 2020, < https://github.com/quankiquanki/skytrax-reviews-dataset/commits?author=quankiquanki>

References

Nguyen, Q 2015, Skytrax User Reviews Dataset (August 2nd, 2015), electronic dataset, quankiquanki, viewed 01 December 2020, < https://github.com/quankiquanki/skytrax-reviews-dataset/commits?author=quankiquanki>