Renewable Energy in France
Safa EL AZRAK - Charles POULLE - Ekaterina MAZANCHENKO
Summary 01 Analysis
We are going to present the story drawn by our analysis
02 Machine Learning
We created a Machine Learning Model with the data we chose
03 Dashboard
To make it more readable, we have imputed all of this within a dashboard
Meet
Our
Team
Ekaterina Mazanchenko Safa El Azrak Charles Poulle
Data Analyst Data Analyst Data Analyst
1. Analysis
Data from 2014 until 2020
The Table 13 Regions
The Coordinates
Nuclear in France
About 40% of missing values on first check but they are not really missing!
Production 5_prod_region.htm
Content Here
03
Content Here
05
Content Here
04
All renewables (wind, solar, hydro, biomass) Wind Solar
2014 20,3% 3,9% 0,3%
2020 26,1% 8,6% 0,7%
Key Regions
6_instal_region.htm
01
Eolian
02
Solar
Instal.htm
Efficiency FC.htm TC.htm
60.5
39.8
Eolian
33.4
18.4
Solar
FC Average
TC Average
What about consumption?
2. Machine
Learning
Preprocessing
The Region Date
Group the Regions
Random Forest (Solar)
Take only Missing
concerned Data
cols
Make Random Forest
17.08 Logistic Regression
Mean
Solar
Create
Target on Split Train
median &Test
And choose
input for
regression
Standardize the data
17.08 Logistic Regression
Mean
Solar
Regression Odd Ratio
Score
Intercept and Coef
Testing the model
0 1
The model will predict
excessive solar
production in:
78.78%
of cases Test
Regression
Score
3. Plotly
Dashboard
Structure of the Dashboard
SUMMARY of the App
• Home page Libraries and packages used in the Dashboard :
• Page 1
- Energy map
• Filter by dropdown menu to choose the input data : • Seaborn palette
•
solar / wind energy, production, consumption
Filter by year slicer
• Dash
• Page 2 • Dash components : core, html, bootstrap
- Energy production and consumption by region • Dash dependencies (input/output)
• Filter by year slicer
• Plotly express
- Comparison of load factors
• Filter by year slicer • Pandas
• Page 3 • Plotly subplots
- Load factors compared to weather (solar / wind) • Plotly io
• Filter by year slicer
• Page 4
- Data sources and team members
Based on a stylesheet style using a Sidebar to navigate through the five pages of the Dash ApplicationC
Intuitive and user friendly, giving the opportunity to filter the data by the variables to observe or by date
or region
Conclusion
• pandas, plotly, dash, datetime, sklearn libraries,
• multi-source analysis,
• ML,
• dash application,
• team work.
Data sources:
• https://opendata.reseaux-energies.fr/explore/dataset/parc-regional-annuel-prod-eolien
-solaire/information/?disjunctive.region
• https://opendata.reseaux-energies.fr/explore/dataset/production-regionale-mensuelle-f
iliere/export/?disjunctive.region
• https://opendata.reseaux-energies.fr/explore/dataset/equilibre-regional-mensuel-prod-
conso-brute/information/?disjunctive.region
Thank You
For listening
