cwawire/Airline-Ticket-Price-Prediction

African Airline Ticket Price Prediction (End-to-End ML) # African Airline Ticket Price Prediction (End-to-End ML) --- ## Overview This project builds a full machine learning pipeline to predict airline ticket prices for major African routes. It simulates a real-world pricing problem where airlines and travel platforms need to estimate ticket prices based on multiple factors such as route, timing, and demand patterns. The goal is to: - Predict airline ticket prices using machine learning - Understand key drivers of price variation - Demonstrate an end-to-end ML workflow from data generation to deployment --- ## Problem Statement Airline ticket prices are highly dynamic and influenced by multiple factors including: - Route and distance - Booking timing - Airline type - Demand fluctuations Traditional pricing methods can be inconsistent and difficult to scale. This project builds a machine learning model to support: - Price prediction - Better pricing insights - Data-driven decision-making --- ## Dataset A synthetic dataset was generated to simulate airline ticket pricing in an African context. Features include: - **airline** — Airline operator - **source_city** — Departure location - **destination_city** — Arrival location - **departure_time** — Time of departure - **stops** — Number of stops - **flight_duration** — Duration of flight - **days_left** — Days before departure - **price** — Target variable --- ## Methodology ### 1. Data Generation - Simulated realistic airline pricing data - Incorporated relationships between time, distance, and demand --- ### 2. Exploratory Data Analysis (EDA) - Analyzed price distribution - Examined relationships between features and ticket price - Identified patterns across routes and booking windows --- ### 3. Data Preprocessing - Handled categorical variables using encoding - Prepared dataset for model training - Split into training and testing sets --- ### 4. Model Development A **Random Forest Regressor** was used to predict ticket prices. Why this model: - H …