s-themis / car-fuel-cost-prediction

Predicting car fuel cost with machine-learning.

Car-Fuel-Cost-Prediction

This is a regression problem, where given various vehicle characteristics we predict the fuel cost at 12,000 miles.

Dataset

The dataset is taken from: https://github.com/amercader/car-fuel-and-emissions

Modelling

• Manual feature selection through data exploration with Pandas.
• Handling of missing values with imputation.
• Scaling numerical features and one-hot encoding categorical features.
• Composition of imputation, preprocessing and regression as a single pipeline with scikit-learn Pipelines and ColumnTransformer.
• Model selection.

How to run the notebook.

Create and activate a virtual environment.

$ python -m venv venv
$ source venv/bin/activate

Install the dependencies.

$ pip install -r requirements.txt

Launch a JupyterLab server.

$ jupyter-lab