Predicting Interview Engagement in Real-Time Recruitment: Class Imbalance and Behavioral Feature Analysis
Bibliographic record
Abstract
Real-time recruitment platforms with bidirectional matching systems pose unique challenges in predicting candidate interview engagement, with only about 3 % of candidates reaching the waiting room stage. Unlike traditional recruitment prediction focusing on static resume-job matching, this study analyzes behavioral dynamics in live interview participation using 73,000 real-world candidate-job matching instances from a North American recruitment platform. We systematically evaluated 120 model configurations that combine five machine learning algorithms with multiple imbalance handling strategies. Our results demonstrate that LightGBM with a <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$1: 3$</tex> class weighting scheme and Tomek Links undersampling achieves the most stable performance across validation experiments <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(\mathbf{F 1}=\mathbf{0. 5 9 7} \pm \mathbf{0. 0 0 4}, \mathbf{9 5 \%}$</tex> CI: 0.585-0.609) with high test performance (Macro F1 <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$=\mathbf{0. 7 9 7 1}$</tex>). Compared to the majority-class baseline classifier (Macro F1 = 0.4926), this corresponds to an absolute improvement of 0.3045. In addition, feature analysis highlights that user engagement metrics and temporal factors are as predictive as algorithmic match scores, with class weighting consistently outperforming aggressive resampling strategies. The study uncovers counterintuitive behavioral patterns, including a negative correlation between resume-job match scores and interview participation, suggesting algorithmic refinement opportunities. These findings provide actionable insights for the optimization of the recruitment platform and establish a methodological framework applicable to other real-time matching systems with extreme class imbalance.
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 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 | 0.002 | 0.000 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.002 | 0.003 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.000 | 0.001 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 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 teacher head, 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".