Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/s-matke/eco-forecast
Machine learning model used for predicting European country with most green surplus energy generated
https://github.com/s-matke/eco-forecast
data-science green-energy machine-learning scikit-learn supervised-learning
Last synced: about 1 month ago
JSON representation
Machine learning model used for predicting European country with most green surplus energy generated
- Host: GitHub
- URL: https://github.com/s-matke/eco-forecast
- Owner: s-matke
- Created: 2023-11-18T09:12:24.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2023-11-21T16:19:14.000Z (about 1 year ago)
- Last Synced: 2024-10-22T00:57:32.554Z (3 months ago)
- Topics: data-science, green-energy, machine-learning, scikit-learn, supervised-learning
- Language: Jupyter Notebook
- Homepage:
- Size: 1.48 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Schneider Electric Europe Data Challenge
## Overview
With the increasing digitalization and the growing reliance on data servers, Schneider Electric presents an innovative challenge to predict the European country with the highest surplus of green energy in the next hour. This prediction is crucial for optimizing computing tasks, utilizing green energy effectively, and reducing CO2 emissions.
## Objective
Create a model to predict the European country with the highest surplus of green energy in the next hour. Consider energy generation from renewable sources and energy consumption. The solution should align with Schneider Electric's ethos and present an unprecedented approach.
## Dataset
Utilize time-series data from the ENTSO-E Transparency portal API, including electricity consumption, wind energy generation, solar energy generation, and other green energy generation. Homogenize the data to 1-hour intervals, and create 'train.csv' and 'test.csv' datasets from 01-01-2022 to 01-01-2023.
API Token: `1d9cd4bd-f8aa-476c-8cc1-3442dc91506d` (or alternative tokens provided)
## Repository Structure
```plaintext
|__README.md
|__requirements.txt
|
|__data
| |__train.csv
| |__test.csv
|
|__src
| |__data_ingestion.py
| |__data_processing.py
| |__model_training.py
| |__model_prediction.py
| |__utils.py
|
|__models
| |__model.pkl
|
|__scripts
| |__run_pipeline.sh
|
|__predictions
|__predictions.json
```
## Installation and Usage
1. **Clone the Repository:**
```bash
git clone https://github.com/s-matke/eco-forecast.git
cd eco-forecast
```
2. **Install Dependencies:**
```bash
pip install -r requirements.txt
```
3. **Run Pipeline:**
```bash
./run_pipeline.sh
```
For example:
```bash
./run_pipeline.sh 2022-01-01 2023-01-01 data/raw_data.csv data/processed_data.csv models/model.pkl data/test_data.csv predictions/predictions.json
```
## Data Processing
- Missing values are imputed as the mean between preceding and following values.
- Data with resolution finer than 1 hour is resampled to an hourly level.
- Non-green energy sources are discarded.
- The resulting CSV includes columns per country representing generated green energy per energy type and load.
- Calculated surplus as the difference between generated green energy and load in order to create label feature
- Downsampled labels that are dominating, however, it's set as an optional parameter
## Model
Used Multi-layer Perceptron (MLP) classifier as our model from scikit-learn module to predict the country with the highest surplus green energy. The model is saved inside models directory while it's predictions on test dataset have been saved in predictions directory.