Weather Predictions with Machine Learning

About

I am looking at various ways to use ML with weather patterns. This repo was created by following a tutorial: Predict the Weather with Machine Learning by Vik Paruchri with Dataquest.

Model

$$\sum_{i=1}^n (y_i - \sum_{j=1}^p x_{ij}\beta_j)^2 + \lambda \sum_{j=1}^p \beta_j^2$$

Data

Data was requested through the National Oceanic and Atmospheric Administration and National Centers for Environmental Information's Climate Search request pages. Parameters of search requested were "Daily Summaries" taken from the Louisville Muhammad Ali Airport (SDF) from January 1, 1960 to June 28, 2022.

The five core weather patterns are established from the documentation¹ and include the following values per day:

1. Precipitation in inches
2. Snowfall total in inches
3. Snow depth (ground accumulation) in inches
4. High temperature in Fahrenheit
5. Low temperature in Fahrenheit

Running the Program

Requirements

Anaconda is the preferred distribution and contains all the required packages needed to run this program.

If you would rather install individually:

pip install jupyter
pip install pandas
pip install -U scikit-learn
pip install matplotplib
pip install numpy

Steps

1. Clone or download the repo
2. In the root folder of the repo, install requirements (if not using Anaconda).
3. Open the weather.ipynb notebook with jupyter notebook in the terminal, or open with editor.
4. Click "Run All"

Clear the kernel and restart the notebook as needed.

Footnotes

Footnotes

1. See GHCND_documentation.pdf for the Global Historical Climatology Network. ↩