Gradient boosting and Shapley additive explanations for fraud detection in electricity distribution grids
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Fraud in electrical energy consumption represents a critical economic burden for utility companies around the world. Despite systematic efforts to mitigate electricity theft, this practice persists mostly in developing countries where companies rely on traditional detection methods. In Brazil it is estimated that around 7% of the total electrical energy available for consumption in 2016 was lost due to frauds. Here we describe an efficient and scalable system to predict fraudulent behavior and guide in loco inspections. We compared the performances of several machine learning algorithms using consumption and inspection data provided by CPFL Energia. We show that proper feature engineering and boosted classification trees trained with XGBoost are able to extract patterns related to fraud occurrence and to achieve predictive power of practical consequences. Moreover, we demonstrate how Shapley additive explanation (SHAP) values can be employed to build user friendly explanations. Together, the proposed model and its explainers contribute not only to reveal potentially fraudulent behavior but also to understand root causes, what can be used to devise robust mitigation strategies.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it