rudyvdbrink/Tunis_Air_prediction

Flight delay prediction # Tunis Air flight delay prediction Machine learning challenge to predict flight delays for Tunis Air. Information on the challenge can be found here. Before running the code, you need to download the data. For this coding challenge I used some of the code from this repository. And overview of some of the the models I've tried: The best solution to the coding challenge (that I could find) involved breaking the problem down into two distinct components: one a classification task, and one a regression task. The classification task involved predicting if a flight is delayed (or not), and the regression task involved predicting the amount of delay of a flight, assuming that it is delayed. ### List of files: - `0_EDA.ipynb`: Initial data exploration and making some plots. - `1_preprocess_data.ipynb`: Prepare data for ML models (including train-test split). - `2a_fit_classifcation_model.ipynb`: Fit XGBoost classifier to make binary delay / on time predictions. - `2b_fit_regression_model.ipynb`: Fit XGBoost regression model to make graded delay predictions. - `3_combine_models.ipynb`: Make a predictions based on the combination of the classifier and regressor, and run model evaluation. ### **Installation, for `macOS`** do the following: - Install the virtual environment and the required packages: ```BASH pyenv local 3.11.3 python -m venv .venv source .venv/bin/activate pip install --upgrade pip pip install -r requirements.txt ``` ### **Installation, for `WindowsOS`** do the following: - Install the virtual environment and the required packages: For `PowerShell` CLI : ```PowerShell pyenv local 3.11.3 python -m venv .venv .venv\Scripts\Activate.ps1 python -m pip install --upgrade pip pip install -r requirements.txt ``` For `Git-Bash` CLI : ```BASH pyenv local 3.11.3 python -m venv .venv source .venv/Scripts/activate python -m pip install --upgrade pip pip install -r requirements.txt ```