Hybrid Rank Aggregation (HRA): A novel rank aggregation method for ensemble-based feature selection
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
Abstract Background Feature selection (FS) reduces the dimensions of high dimensional data. Among many FS approaches, ensemble-based feature selection (EFS) is one of the commonly used approaches. The rank aggregation (RA) step influences the feature selection of EFS. Currently, the EFS approach relies on using a single RA algorithm to pool feature performance and select features. However, a single RA algorithm may not always give optimal performance across all datasets. Method and Results This study proposes a novel hybrid rank aggregation (HRA) method to perform the RA step in EFS which allows the selection of features based on their importance across different RA techniques. The approach allows creation of a RA matrix which contains feature performance or importance in each RA technique followed by an unsupervised learning-based selection of features based on their performance/importance in RA matrix. The algorithm is tested under different simulation scenarios for continuous outcomes and several real data studies for continuous, binary and time to event outcomes and compared with existing RA methods. The study found that HRA provided a better or at par robust performance as compared to existing RA methods in terms of feature selection and predictive performance of the model. Conclusion HRA is an improvement to current single RA based EFS approaches with better and robust performance. The consistent performance in continuous, categorical and time to event outcomes suggest the wide applicability of this method. While the current study limits the testing of HRA on cross-sectional data with input features of a continuous distribution, it could be applied to longitudinal and categorical data.
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 | 0.003 | 0.004 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.000 |
| 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 it