Weather Forecasts Batch Data Pipeline

BigQuery pipeline is an implementation of an end-to-end batch data pipeline which runs in a weekly manner. It ingests the latest weather forecasts data from MetaWeather API, loads that data to Google BigQuery and finally produces weather insights via a dashboard.

The pipeline was built using Python; Pandas; BigQuery API; Heroku CLI; Google Data Studio

Table of Contents

• Architecture diagram
• How it works
  • Data pipeline
  • Additional work
• Running project
• Final results
  • Dataset
  • Dashboard
• References

Architecture diagram

How it works

Data pipeline ↗️

• Extract: call location endpoint to get the current weather of a city and 5 days forecast
• Transform: do some transformations like: renaming columns, changing data types, and generate a new date dimension table from the date column
• Load: load the final weather data and the generated dimension to bigquery

Additional work

• Historical data: get history of weaher forecasts using location day endpoint between two selected dates
• API cities: get the cities available to query from MetaWeather API using loaction search endpoint
• Logging: implement simple custom logger using loguru libirary
• Scheduling: run the pipeline in a weekly manner as the MetaWeather API provides forecasts for 6 days interval

Running project

First, you should setup your google service account permission and create the required tables using BigQuery UI

Then clone this repo and run the pipeline

pip install -r requirements.txt

python main.py

Final results

Dataset

Dashboard

References

• BigQuery API Client Libraries
• Heroku google application credentials buildpack
• Scheduled Jobs with Custom Clock Processes in Python