Generate Data Branch Tech Challenge – README

Overview

This project analyzes international passenger traffic data for AeroConnect to identify high- and low-demand routes, uncover seasonal and regional trends, and build a model to forecast traffic for resource allocation decisions.

The challenge required:

1. Cleaning and preparing the dataset.
2. Performing exploratory data analysis (EDA).
3. Building a predictive model.
4. Evaluating the model’s accuracy.
5. Providing recommendations for AeroConnect’s future route planning

Data Cleaning

I cleaned the dataset by:

1. Converting Year and Month_num into a proper Date column.
2. Checking for missing or inconsistent values.
3. Creating a Route column.
4. Sorting the dataset chronologically for modeling.

Cleaned data link: https://drive.google.com/file/d/17vA1u8qtmHL24LFEBr2MuxxVBUZV_oqm/view?usp=sharing

Process

1. Setup: Downloaded and installed all necessary Python libraries (e.g., pandas, numpy, matplotlib, scikit-learn) to support data cleaning, visualization, and modeling.
2. EDA: Identified the most and least trafficked routes, seasonal travel patterns, and geographical concentration of international traffic.
3. Model: Built a regression model with a third-degree polynomial trend, monthly dummy variables, and ridge regression (α = 0.1) to capture both long-term growth and seasonality.
4. Evaluation: Assessed accuracy using MAE, RMSE, and MAPE.
5. Recommendations: Suggested which routes AeroConnect should expand or scale back, and how to apply forecasts for planning.