This project analyses Transmission Loss Adjustment Factors (TLAFs) in the Irish Single Electricity Market (SEM) using data analysis, clustering, geospatial mapping, and explainability techniques. The goal is to identify patterns in generator transmission loss behaviour and group generators with similar TLAF characteristics.
The project combines TLAF data with mapped power plant and infrastructure information, then applies clustering models to explore whether meaningful structures exist in the data. It also produces visual outputs such as maps, elbow plots, dendrograms, SHAP summaries, and cluster-level comparisons.
The main aim of this project is to investigate whether machine learning and spatial analysis can be used to better understand transmission efficiency across generators in Ireland. By clustering generators based on TLAF-related and location-linked features, the project attempts to identify high-loss and low-loss patterns and provide useful insight for grid analysis.
- Cleaning and compiling raw TLAF and plant data
- Mapping plants to Irish geospatial boundaries
- Creating matched and unmatched plant outputs
- Building GeoJSON outputs for visualisation
- Applying clustering models such as:
- K-Means
- DBSCAN
- Hierarchical Agglomerative Clustering
- Gaussian Mixture Models
- Comparing clustering performance visually and analytically
- Generating SHAP-based explainability outputs
- Producing map-based visualisations of plants and TLAF patterns
TLAF Final Project/
│
├── compiled_data/ # Processed and merged datasets used in analysis
├── raw_data/ # Original source files and raw inputs
├── output/ # Generated maps, plots, SHAP visuals, and clustering outputs
│
├── main_part_1.ipynb # Main notebook for analysis, clustering, and visual outputs
├── main_part_2.ipynb # Updated notebook for analysis, clustering, and visual outputs
│
└── README.md # Project documentation