✈️ Customer Satisfaction Probability Prediction

A machine learning web application that predicts whether an airline passenger is satisfied or neutral/dissatisfied based on their flight experience and service ratings.

---

📌 Table of Contents

• Overview
• Demo
• Project Structure
• Dataset
• Exploratory Data Analysis
• Model
• Features
• Installation
• Usage
• Model Performance
• Tech Stack
• Author

---

Overview

This project builds an end-to-end machine learning pipeline to predict customer satisfaction for airline passengers. It covers the full workflow — from raw data exploration and preprocessing to model training, evaluation, and deployment as an interactive Streamlit web application.

The trained model takes 22 passenger-related features (demographics, travel type, flight class, service ratings, and delays) and outputs a binary prediction: Satisfied 😆 or Neutral / Dissatisfied 😒.

---

Demo

Run locally using the instructions in the Usage section.

Sidebar Inputs	Prediction Output
Demographics, travel class, service ratings	✅ Satisfied / ❌ Neutral or Dissatisfied

---

Project Structure

├── data_preprocessing_and_model_training.ipynb   # Full EDA, preprocessing & model training notebook
├── app.py                                         # Streamlit web application
├── model.pkl                                      # Serialized trained Random Forest model
├── customer_satisfication_dataset.csv             # Raw dataset (not included — see Dataset section)
├── requirements.txt                               # Python dependencies
└── README.md                                      # Project documentation

---

Dataset

The dataset used is the Airline Passenger Satisfaction dataset, containing survey responses and flight details for thousands of passengers.

Key columns include:

Column	Type	Description
Gender	Categorical	Male / Female
Customer Type	Categorical	Loyal / Disloyal Customer
Age	Numeric	Passenger age
Type of Travel	Categorical	Business / Personal Travel
Class	Categorical	Business, Eco, Eco Plus
Flight Distance	Numeric	Distance of the flight (miles)
Inflight wifi service	Numeric	Rating 0–5
Seat comfort	Numeric	Rating 0–5
Departure Delay in Minutes	Numeric	Departure delay duration
Arrival Delay in Minutes	Numeric	Arrival delay duration
satisfaction	Target	Satisfied / Neutral or Dissatisfied

Note: The dataset file (customer_satisfication_dataset.csv) is not included in this repository. You can find a compatible dataset on Kaggle — Airline Passenger Satisfaction.

---

Exploratory Data Analysis

The notebook covers a thorough EDA including:

• ✅ Shape, dtypes, and column inventory
• ✅ Missing value analysis (Arrival Delay in Minutes — filled with median)
• ✅ Duplicate detection
• ✅ Satisfaction distribution (count plot)
• ✅ Satisfied vs. Dissatisfied breakdown by Gender, Class, and Type of Travel
• ✅ Correlation heatmap across all numeric features

---

Model

A Random Forest Classifier was trained with the following configuration:

RandomForestClassifier(n_estimators=350, min_samples_split=11)

Preprocessing steps applied before training:

1. Dropped irrelevant columns: Unnamed: 0, id
2. Imputed missing values in Arrival Delay in Minutes with the column median
3. Label-encoded all categorical features:
   • Gender: Male → 1, Female → 0
   • Customer Type: Loyal → 1, Disloyal → 0
   • Type of Travel: Business → 1, Personal → 0
   • Class: Business → 1, Eco → 2, Eco Plus → 3
   • satisfaction: Satisfied → 1, Neutral/Dissatisfied → 0
4. 80/20 train-test split (random_state=42)

The trained model is serialized and saved as model.pkl using Python's pickle module.

---

Features

The Streamlit app accepts the following 22 input features via the sidebar:

#	Feature	Input Type	Range / Options
1	Gender	Selectbox	Male, Female
2	Customer Type	Selectbox	Loyal, Disloyal
3	Age	Number Input	7 – 85
4	Type of Travel	Selectbox	Business, Personal
5	Class	Selectbox	Business, Eco, Eco Plus
6	Flight Distance	Number Input	31 – 4983
7	Inflight wifi service	Number Input	0 – 5
8	Departure/Arrival time convenient	Number Input	0 – 5
9	Ease of Online booking	Number Input	0 – 5
10	Gate location	Number Input	1 – 5
11	Food and drink	Number Input	0 – 5
12	Online boarding	Number Input	0 – 5
13	Seat comfort	Number Input	0 – 5
14	Inflight entertainment	Number Input	0 – 5
15	On-board service	Number Input	0 – 5
16	Leg room service	Number Input	0 – 5
17	Baggage handling	Number Input	1 – 5
18	Checkin service	Number Input	0 – 5
19	Inflight service	Number Input	0 – 5
20	Cleanliness	Number Input	0 – 5
21	Departure Delay in Minutes	Number Input	0 – 1592
22	Arrival Delay in Minutes	Number Input	0 – 1584

---

Installation

1. Clone the repository

git clone https://github.com/zakir-maswani/airline-satisfication-prediction.git
cd customer-satisfaction-prediction

2. Create a virtual environment (recommended)

python -m venv venv
source venv/bin/activate       # On macOS/Linux
venv\Scripts\activate          # On Windows

3. Install dependencies

pip install -r requirements.txt

---

Usage

Run the Streamlit App

streamlit run app.py

Then open http://localhost:8501 in your browser.

Retrain the Model

Open and run the Jupyter notebook:

jupyter notebook data_preprocessing_and_model_training.ipynb

This will regenerate model.pkl with the latest data.

---

Model Performance

Metric	Score
Accuracy	96%
Precision	95%
Recall	98%

Evaluated on a held-out 20% test set.

---

Tech Stack

Category	Library / Tool
Language	Python 3.x
Data Manipulation	pandas
Visualization	matplotlib, seaborn
Machine Learning	scikit-learn
Model Serialization	pickle
Web App	Streamlit
Notebook	Jupyter

---

Author

Built by Zakir Ali — Machine Learning Engineer

---

⭐ If you found this project helpful, feel free to star the repository!