Using TensorFlow and Long Short-Term Memory Recurrent Neural Networks.
Background: research has shown that temperature, humidity, and wind can be accurately predicted out to 72hrs using a 15 year data period of hourly measurements. Mostly based on this paper.
Set up a Python virtual environment and install requirements via pip:
$ virtualenv venv
$ source venv/bin/activate
$ pip install -r requirements.txt
Scripts:
To download data form Weather Company Data, use tools/weather_data.py > data.csv. This will download the last year's daily minimum and maximum temperature values for a selected location and save it in data.csv.
Before use, make sure to sign up for an account with them and add the username and password to the .env file.
Richard Dancsi
- Blog: wimagguc.com
- Github: github.com/wimagguc
- Twitter: twitter.com/wimagguc
- Linkedin: linkedin.com/in/richarddancsi
- Company: buntlabs.com