ruankie / weather-report

Creates a five-day temperature forecast for any group of cities. Forecasts are generated by parameterised jupyter notebooks that run in parallel.

Details

This repository is an updated version of this tutorial on Towards Data Science (for September 2021) that produces a five-day temperature forecast for different cities. The following changes were made:

1. Incorporates new API calls - the older ones no loger work.
2. Incorporates parallel processing so that forecasts can be made of different cities at the same time.
3. Does not produce .pdf report files for forecasts - graphs produced in .png image format instead.

The template notebook is used to forecast temperature for the next five days. In each slave notebook, the relevant city name is injected as a parameter so notebooks can be executed independently and in parallel. This also allows each slave notebook to function as a kind of report for a particular city where all the code and results are contained in the same file.

This repository can be used for learning how to use the papermill library that is used to parameterise jupyter notebooks. It can als be used a guide to see how different parameterised notebooks can be executed in parallel using the multiprocessing library for more efficient use of computer resources.

Setup and Execution

1. This repository uses some specific packages to forecast weather (pyowm) and parameterise notebooks (papermill). Create a separate virtual environment with all the relevant packages for this project by running the following from your terminal:

conda create -n papermill-weather python=3.7
conda activate papermill-weather
pip install -r requirements.txt

2. You will also have to create your own profile and generate a free API key on the Open Weather Map website.

3. When your environment and API keys are set up as in step 1. and 2., you can proceed to choose the cities you want temperature forecasts for by filling the dictionary in the forecast_template.ipynb notebook file. Give each city a name along with its longitude and latitude coordinates in decimal format.

4. Assign your own API key that you created in step 2. to the api_key variable inside the forecast_template.ipynb notebook file OR create a file called my_api_key.py in the root directory of this repository and create a variable called api_key inside it with the value of your API key assigned to it. For the second option, you will need to import your API key from your my_api_key.py file into the forecast_template.ipynb notebook.

5. Finally, update the cities list inside parallel_forecast.py file with your city names and start the parallel forecasts from you terminal by running the following command:

python parallel_forecast.py

Results

Your temperature forecasts (min, max, and day average) will be saved to plots in the figures folder.
The slave notebooks will be contained in the slave_notebooks folder.