Predicting Customer Satisfaction

This report focuses on analyzing airline passenger satisfaction using data mining techniques. It utilizes a dataset from a corporate survey involving 129,880 customers, examining various factors such as gender, age, travel type, and service satisfaction levels. The report discusses data preprocessing, feature selection, and the application of predictive models, specifically Gaussian Naive Bayes and Logistic Regression. It evaluates model performance through accuracy and ROC-AUC scores, concluding that both models perform well, with Logistic Regression slightly outperforming Gaussian Naive Bayes in predicting customer satisfaction.

Table of Contents

• Data Contents
• Data Understanding
• Data Mining and Prediction Design
• Models Used
• Evaluation Metrics
• Conclusion

Data Contents

The dataset consists of two files: train.csv and test.csv, which contain demographic and satisfaction-related information from 129,880 airline passengers. Key features include:

Feature	Description
Gender	Passenger gender (Female, Male)
Customer Type	Type of customer (Loyal, Disloyal)
Age	Passenger age
Type of Travel	Flight purpose (Personal, Business)
Class	Travel class (Business, Economy, Economy Plus)
Flight Distance	Distance of the flight
Inflight Wifi Service	Wifi satisfaction (0: Not applicable; 1-5)
Departure/Arrival Time	Convenience of departure/arrival time satisfaction
Ease of Online Booking	Online booking satisfaction
Gate Location	Gate location satisfaction
Food and Drink	Food and drink satisfaction
Online Boarding	Online boarding satisfaction
Seat Comfort	Seat comfort satisfaction
Inflight Entertainment	Entertainment satisfaction
On-board Service	Onboard service satisfaction
Leg Room Service	Legroom service satisfaction
Baggage Handling	Baggage handling satisfaction
Check-in Service	Check-in service satisfaction
Inflight Service	Inflight service satisfaction
Cleanliness	Cleanliness satisfaction
Departure Delay in Minutes	Minutes of departure delay
Arrival Delay in Minutes	Minutes of arrival delay
Satisfaction	Overall satisfaction (Satisfied, Neutral, Dissatisfied)

Data Understanding

The data is processed using Python libraries such as Pandas and NumPy. Initial steps include:

• Exploratory Data Analysis (EDA)
• Data cleaning and transformation
• Handling missing values

Data Mining and Prediction Design

The project employs data mining techniques to uncover relationships between features and customer satisfaction. The following methodologies are utilized:

1. Correlation Analysis
2. Data Visualization
3. Principal Component Analysis (PCA)

Models Used

Two predictive models are implemented:

1. Gaussian Naive Bayes (GNB)
2. Logistic Regression

Both models are trained on the processed training dataset and evaluated on the test dataset.

Evaluation Metrics

The performance of the models is assessed using:

• Accuracy Score
• ROC-AUC Score
• Classification Report
• Confusion Matrix
• ROC-AUC Curve

Conclusion

The Logistic Regression model slightly outperforms the Gaussian Naive Bayes model in predicting customer satisfaction. Key features influencing satisfaction include the type of travel and customer loyalty status.

Getting Started

To run this project, ensure you have the following libraries installed:

• pandas
• numpy
• matplotlib
• seaborn
• scikit-learn

Clone the repository and run the Jupyter Notebook to explore the analysis and predictions.

License

This project is licensed under the MIT License - see the LICENSE file for details.