Novel Metrics for Evaluation and Validation of Regression-based Supervised Learning
Bibliographic record
Abstract
Error consistency is a validation metric for evaluating the sample-based error variability across machine learning models trained as part of in-lab validation. Many machine learning (ML) based regression algorithms are likely to be inconsistent with each other when trained repeatedly on the same task as part of standard cross validation, in part due to sampling, but also, potentially associated with the inclusion of randomness in their training paradigms, which is common in many learning techniques. In this work, we propose a novel approach to validation and evaluation of regression-based learning algorithms, called regression ‘error consistency’ (EC), to assist in assessing sample-wise consistency of errors as part of in-lab validation. We have applied novel EC metrics on six real-world datasets with six different regressors, evaluated the model performance with well-known metrics and compared the results with previously developed classification EC. The results demonstrate that, out of six models, the random forest achieved high accuracy but exhibited less consistency in its error profiles. This finding matches with classification based EC results. In addition, we applied the EC metrics on the MNIST digits dataset using a convolutional neural network (CNN) as part of a preliminary deep learning experiment. Though MNIST is typically treated as a classification dataset, we considered this dataset as a regression problem and the CNN model developed demonstrated good performance. We believe that the proposed EC metrics will be useful in evaluating and validating regression algorithm error consistency, including in deep learning, and will hopefully guide the machine learning research community to develop more reproducible and predictable (in terms of the errors they will make) regression algorithms. Public domain code is provided.
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 | 0.047 | 0.171 |
| Meta-epidemiology (narrow) | 0.003 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.005 | 0.003 |
| Science and technology studies | 0.001 | 0.003 |
| Scholarly communication | 0.003 | 0.004 |
| Open science | 0.003 | 0.003 |
| Research integrity | 0.003 | 0.004 |
| 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".