Weather Prediction MLOps

This project is part of the "Cloud Computing" course (RSO) and focuses on Machine Learning in the cloud for weather prediction.

Project Structure

The project includes the following components:

• dags: Apache Airflow DAGs for ETL (Extract, Transform, Load), evaluation, and training.

  • etl.py: ETL DAG for data processing.
  • eval.py: Evaluation DAG for model evaluation.
  • train.py: Training DAG for model training.

• data: Data files obtained from the Open-Meteo weather API.

• models: Trained machine learning models in .h5 format. The .best extension indicates the best-performing model.

• open-meteo-api.ipynb: A Jupyter Notebook containing code for fetching, processing, and loading data into BigQuery from the Open-Meteo API.

• weather-pred-nn.ipynb: A Jupyter Notebook that includes exploratory data analysis and training different types of machine learning models (naive, autoregressor, neural network).

Setting Up Apache Airflow

To set up Apache Airflow for this project, follow these steps:

1. Create a virtual environment and activate it:

   python3 -m venv airflow-venv
   source airflow-venv/bin/activate

2. Install the required Python packages from requirements.txt:

   pip install -r requirements.txt

3. Initialize the Airflow database:

   airflow db init

4. Create an admin user for Airflow:

   airflow users create \
       --username admin \
       --firstname FIRST_NAME \
       --lastname LAST_NAME \
       --role Admin \
       --email [email protected]

5. Set the Google Cloud service account key file path as an environment variable:

   export GOOGLE_APPLICATION_CREDENTIALS="/path/to/service-account-key.json"

6. Start the Airflow webserver on port 8080:

   airflow webserver -p 8080

7. Start the Airflow scheduler:

   airflow scheduler -S .

8. If needed, you can forcefully kill the Airflow webserver and its processes on port 8081:

   sudo fuser -k 8081/tcp

9. Important: Before running the DAGs, set the necessary Airflow variables via the Airflow Web UI. Specifically, add the following variable:

   gcp_info = {
       "project_id": "balmy-apogee-404909",
       "bucket_name": "europe-central2-rso-ml-airf-05c3abe0-bucket",
       "dataset_id": "weather_prediction",
       "weather_table_id": "weather_history_LJ",
       "predict_table_id": "weather_predictions",
       "model_table_id": "weather_models"
   }

   This variable is essential for the correct functioning of the DAGs, as it provides necessary configuration information.

Now, you should have Apache Airflow set up and running for your project.