Predicting Electricity Demand in the Energy Grid

Guided Mode Initialization

AUploading Dataset

Click on the **'Upload CSV or Excel Data'** button → Select a source for the dataset → Upload `electricity_demand_train.csv`. The system automatically analyzes the file, converting it into feature data, extracts column descriptions, and identifies the top value-adding targets for prediction.

BChoosing Analysis Mode

Operation Using Autonomous Guided Mode

AQuery Response

The factors impacting electricity demand in England and Wales include the settlement period, network demand (nd), total system demand (tsd), and various forms of embedded generation such as wind and solar. In the conducted analysis, machine learning methods were employed yielding estimates of approximately 23,921.29 MW (Linear Regression).

The Linear Regression model shows a very low test error, making it highly reliable for practical use. The Random Forest model should be used with caution, and LightGBM model exhibits a significantly higher error rate.

Although the autonomous guided mode achieves an amazing accuracy of 99.5%, it is likely overfitted to some extent. Manual model building and fine-tuning have to be done to solve this issue.

BAI Application

In automated mode, running the query generates the AI application where users can adjust key variables and see how changes impact the predicted outcome.

Model Fine-Tuning/Manual Model Building

Manual model building allows users to take full control of the model creation process by guiding them step-by-step through data preparation, feature selection, algorithm configuration, model training, and validation.

ASelecting Prediction Target

The 'england_wales_demand' column was selected as the target.

BSelecting Analysis Type

The analysis target is a numeric column. Hence, the 'Regression' analysis type is selected.

CSelecting Model Group/Item

No item/group is required for this dataset (single item).

DSelecting Features

Select the following features: embedded_wind_generation, embedded_wind_capacity, embedded_solar_generation, embedded_solar_capacity, pump_storage_pumping, ifa2_flow, east_west_flow, nsl_flow, eleclink_flow.

ESelecting Training Level

Select the "Moderate" training level. It trains Linear Regression and Xtreme Gradient Boosting with 5-fold cross-validation and 20% test data. Click 'Train'.

AI Modeling Details

The best-performing model was Xtreme Gradient Boosting, with a training error of 3.89% and a testing error of 5.57%. Linear Regression had a higher test error of 13.67%, making it less reliable for real-world applications.

Training Analysis Details

APredicted Target

The scatter plot illustrates predicted vs. actual energy demand values for different models.

BPredicted Trend

The predicted energy demand trends from both models closely followed actual demand values, but some discrepancies were evident.

CError Trend

Linear Regression had a training MAEP of 13.80% and a testing error of 13.67%. Xtreme Gradient Boosting achieved a training error of 3.89% and a test error of 5.57%.

DFeature Importance

Feature importance analysis revealed that variables like 'pump_storage_pumping' and 'embedded_solar_capacity' significantly influenced the models.

Finalize Models

Once satisfied with performance, click 'Deploy'. The system saves and deploys models for future demand analysis or production environment.

Manual Model Building

This AI application enables users to predict electricity demand in England and Wales. Users can manually adjust input sliders for features such as 'pump_storage_pumping,' 'embedded_solar_capacity,' and various interconnector flows to explore different prediction scenarios.

AI Application Demo

1. Adjust key variables using the sliders.
2. Observe how these changes impact the predicted electric demand in real-time.

Saving the Project

Save your project by clicking the icon at the bottom left corner of the textbox.

Sharing the Project

Share the application for single on-demand predictions once the analysis is saved.