In this project we will analyze a dataset which consists of 1094 daily observations related to the wind energy production of a German plant from 01-01-2017 to 12-30-2019.
The original project was made together with 3 university colleagues of mine for the course "Time Series Analysis" (it was implemented in R). The approaches used are:
- a simple Autoregressive model
- an autoregressive model with exogeneous variables (wind capacity and temperature)
The order of the first one is direcly choosen by looking at the autocorrelation (partial), while for the second one the criteria are BIC and AIC.
After the exam, I decided to convert the original R project to Python language and then create by myself a web-app using flask that allows the user to select a specific forecast period to obtain the values from the trained model.
After repo download:
- usual
pip install -r requirements.txtin your environment - then, from terminal
python app.py - copy and paste http://127.0.0.1:5000/predictdata in your browser (
‼️ while runningapp.py‼️ ) - then just put the number of days (integer) from 1 to 59 (because after the 59th day the prediction will be 0)
