Wetterfrosch

Compare different weather websites for your location and find out which one's the best.

How it works

A script schedules a set of crawlers for various weather websites at certain times of the day and fetches data continously. Each crawler produces it's own data file which has to be merged if a full analysis is planned.

How to run

You need to have Pipenv installed. Run the following to install the dependencies and run all crawlers:

# install dependencies
pipenv install
# run the crawlers
pipenv run ./run_all.py

Crawlers

For each weather station, a separate crawler has to be implemented. Each crawler is located in a separate folder in the the crawlers directory (e.g. crawlers/wetteronline). You can use any framework or programming language for the crawlers as long as the following criteria are implemented:

• There is an executabe file called run.* or simply run in the top level directory of the crawler, this is the entry point of the crawler
• The crawler should be able to run without any command line arguments
• The crawler should store it's results in csv files data/{crawler_name}*.csv. You can replace * with everything you want.
• The crawler should create csv files if they don't exist and append if they do exist
• The CSV file should contain the following columns:
  • forecast_time: The start of the timeframe of the forecast
  • forecast_time_end: The end of the timeframe of the forecast
  • temp (or max_temp and min_temp if available): The temperature. If the forcast is for a longer timeframe, there sometimes is a maximum and a minimum temperature available.
  • precip_prob: The precipitation probability
  • curent_time: The time when the request to the website was initiated
• Until further notice, the crawler should fetch the forecasts for Regensburg

Note that although some websites have simple APIs that can be used, we still call the program that fetches the data from APIs a crawler.

Currently, crawlers for the following websites are implemented:

• darksky.net
• wetter.de
• wetteronline.de

TODOs

• Write more crawlers
• Analyse the data
• Gather reference data (what was the actual weather?)