Workshop
Create and deploy an interactive Energy Time Series Forecaster
Thomas Craessaerts
Data Scientist, Elia Group
Thomas is a data scientist at Elia Group, focusing on developing forecasting solutions for the grid development and energy market teams in Belgium. He has a background in electrical engineering and intelligent mechanics.

Session description
In this workshop, we will show you how to build your own auto-Time Series forecaster with an interactive front-end dashboard. The focus will be on energy data, using open source data from Elia Group’s open data initiative.
By the end of the workshop, you will have developed your own auto-ml interactive time series forecaster. You can check it out here: https://share.streamlit.io/chris-elia/ts-forecast/main/main.py
In particular, we will cover:
How time series problems differ from other ML problems
Open Source packages for training time series models
Building an interactive front-end app in Streamlit where users can upload a new time series and get the best prediction model
Deploying your app to the cloud
Structure
Intro on how time series forecasting differentiates from other ML tasks
Getting familiar with Elia OpenData API
Integration of a time series algorithm (using Prophet)
Development of the User Interface with Streamlit
Deployment on Streamlit Cloud
Expected knowledge:
Basic knowledge of Python, querying APIs, and Git
Familiar with pandas/data manipulation
Software:
Python (3.8.13)
IDE (like Visual Code Studio / PyCharm / etc.)
JupyterNotebooks
Anaconda
Git + GitHub account
For the workshop, we are going to use the libraries prophet and streamlit. Please create a virtual environment with the two packages installed before the workshop.