au-data-science-project

This repo contain the source code for the course Data Science Project at Aarhus University.

The source code serves a streamlit application with a dashboard that power suppliers can use to produce their demand forecast.

How to run the app

Requirements:

• Python >= 3.11

Steps:

1. Install dependencies (recommended to use a virtual environment, see guide below): pip install -r requirements.txt
2. Create a file in the root of the project and call it .env - this is used for environment variables
3. Open the .env file and write the following DMI_API_KEY=XXX and replace XXX with an actual API key for the DMI Weather API. An API key can be acquired by subscribing to the metObsAPI from DMI at: https://dmiapi.govcloud.dk/#!/apis/48ed0c1b-ab40-473a-ad0c-1bab40073a51/detail
4. In your terminal go to the directory src/dsp via: cd src/dsp
5. Run ETL flows via: python run_etls.py
6. Now you can run the streamlit app via: streamlit run application.py

Note that running the ETL flows populate the following tables in a local SQLite db (file):

• bronze_consumption
• bronze_prices
• bronze_weather
• silver_consumption
• silver_prices
• silver_weather
• gold_cpw (consumption, prices, and weather data all in one dataset)

Guide (Windows) - Python virtual environment

Steps:

1. Open a terminal in the root of the project
2. Ensure you are on the correct version of Python via: python -V
3. Create virtual environment via: python -m venv .venv
4. Activate the virtual environment via: source .venv/scripts/activate.ps1. Note that you can skip this step in VS Code if you select your .venv as the Python interpreter and then open a new terminal.

Guide (Mac/Linux) - Python virtual environment

Steps:

1. Open a terminal in the root of the project
2. Ensure you are on the correct version of Python via: python3 -V (you can also use a specific version of Python via python3.11 -V)
3. Create virtual environment via: python3 -m venv .venv
4. Activate the virtual environment via: source .venv/bin/activate. Note that you can skip this step in VS Code if you select your .venv as the Python interpreter and then open a new terminal.