Fraud Detection in Utilities Using Data Analytics and Geospatial Analysis
Bibliographic record
Abstract
One of the great challenges of any utilities around the world is the control of losses, which have different causes that are usually classified as technical (accuracy of equipment, leaks and breaks, construction and maintenance procedures) and NON-technical (thefts and frauds), affecting issues such: as investments for the expansion and maintenance of the networks, the profitability of the shareholders and even the continuity of the service.This article proposes a methodology for the detection of NON-technical losses common to any utility company, based on data analytics on business information enriched with data from third parties through geospatial analysis from its geographic location and market segmentation, which allows finding patterns on anomalous situations through supervised (on historical information) and unsupervised (if no information is available) machine learning models.The results of different classification algorithms used in data analytics were analyzed and the one with the highest accuracy and lowest type two error rates (false negatives) was selected to perform field verification work.The methodology was implemented in a natural gas distribution company and was contrasted with methodologies proposed by other authors for electric energy distribution companies, who consider that the problem should be addressed based on an analysis of historical consumption and its deviations.The results obtained with the proposed methodology improve the accuracy and sensitivity of the models by more than 20% and decreases false negatives by the same percentage, facilitating the verification and normalization of customers in anomalous situations and/or fraudulent conditions.
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.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.010 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.009 | 0.007 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.003 | 0.003 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| 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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".