GriP on Data Science

Based on Cookiecutter Data Science, this Data Science project template aims to provide a more opinionated and firm framework, especially tailored towards beginners.

Quickstart

The following should get you set up quickly. For more details, consult the documentation.

Requirements to use the cookiecutter template:

---

• Python>=3.6
• Cookiecutter Python package >= 1.4.0: This can be installed with pip by or conda depending on how you manage your Python packages:

$ pip install cookiecutter

or

$ conda config --add channels conda-forge
$ conda install cookiecutter

To start a new project, run:

---

cookiecutter https://github.com/waveFrontSet/cookiecutter-data-science

Create a new conda environment:

---

make create_environment

Overview over the next steps

---

After activating the conda environment, you are all set up. Here are the next steps:

• Define how to obtain the raw data of your project and update the data/raw target in the Makefile accordingly.
• Define generic processing and clean up transformations in {{ cookiecutter.module_name }}/data/generic_processing.py to produce interim data.
• Define project specific transformations to obtain the final data set in {{ cookiecutter.module_name }}/features/build_features.py.
• Edit {{ cookiecutter.module_name }}/models/model_config.py to decide what models you want to build and what the target value of the prediction will be. Issuing make train will automatically split your dataset into a train and a test set and then fit the models on the train set.
• Edit {{ cookiecutter.module_name }}/models/metric_config.py to decide what metrics you want to use to evaluate the model performance. Issuing make evaluate will evaluate the models using the defined metrics on the test set.

Installing development requirements

---

pip install -r requirements.txt

Running the tests

---

pytest tests