PSXV-29 Late-Breaking: Investigating use of machine learning algorithms to predict days in herd for commercial beef cattle
Bibliographic record
Abstract
Abstract The objective of this study was to explore the potential of Machine Learning (ML) algorithms to increase the accuracy of predicting individual days in the herd (an indicator of stayability) using reproductive records and genomic information. A total of 6943 cows from 3 herds with reproductive performance were included in the study from which 696 cows had genomic information (genotyped using Illumina Bovine 50k SNP BeadChip). Different libraries based on R and Python were used to test various ML models including Lazy Predict, Scikit-learn, PyCaret, and H2O Flow. Genomic information was subjected to quality control by removing SNPs with an allele frequency less than 0.05 or with a call rate lower than 0.95. A total of 42,689 SNP remained for further analysis and accounted for 11% of phenotypic variation (heritability of 0.11±0.02) in DIH. Different numbers of SNPs (500 SNPs, 1K, 5K, 10K, and 15K) were selected based on their contribution to phenotypic variation from GWAS and were included in the models. Model performance measures, such as mean absolute error (MAE) and mean square of error (MSE), worsened with increased SNPs in the model. Bayesian Ridge algorithm using 500 top SNPs contributed to the phenotypic variance, had the best performance to predict DIH with MAE of 612.6 and R2 of 0.52 in the training population using PyCaret program. When BWT and WWT were added to the model, in addition to SNPs, little change was observed in the model’s performance. Overall, we concluded that ML models had better performance compared to the conventional modeling approach and genomic analysis; CatBoost model had 55% lower mean square of error compared to the simple linear regression (734650 vs 1637410). The results suggest that ML tools have the potential to improve the accuracy of predicting DIH compared to simple linear regression and conventional genomic analysis.
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 | 0.004 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| 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 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".