Predicting self-exclusion status in online gambling data via machine learning algorithms
Bibliographic record
Abstract
Abstract: The identification of problematic gambling in online gamblers from behavioural data (‘player tracking’) may enable interventions to be targeted to those users experiencing harms. This study tested the predictive performance of machine learning models in classifying online gamblers based on voluntary self-exclusion (VSE) status as a binary indicator of problem gambling. We used 1 year of de-identified data from the eCasino section of the PlayNow.com platform in British Columbia, Canada, comprising 31,115 users placing over half a billion individual bets. Input variables were based on daily-aggregate and session-aggregate measures capturing gambling frequency, intensity, and variability. To mitigate concerns about the ‘black box’ nature of machine learning, we report ‘feature importance’ values to show the variables that are most predictive. The primary model compared 1323 self-excluders against an under-sampled (n = 3000) control group. Across 6 variants of our machine learning model, we obtained classification performance (AUROC) from 75 to 79%. Variability in a monetary measure of gambling intensity (Variance in Money Bet per Session) showed the highest feature importance value. Model predictions were used to classify control participants in three risk levels based on resemblance to self-excluders; these risk subgroups differed significantly on each of the 9 input variables. Implications: Machine learning can classify online gamblers based on self-exclusion status with 75-79% performance, using relatively coarse input variables that do not require baseline data or analysis of trajectories. Next steps are to establish convergence across different gambling forms, and using alternative markers of problem gambling.
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 | 0.006 | 0.013 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
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".