Statistical Data Validation Methods for Large Cheese Plant Database
Bibliographic record
Abstract
Production data of the cheesemaking process are used to monitor milk fat and protein recoveries in cheese, cheese yield, and composition and eventually to predict these parameters. Due to the large impact of these factors on cheese quality and plant profitability, it is very important to use reliable data for analysis, modeling, and control of the process. This paper tested six methods for detecting erroneous data in industrial cheesemaking databases. The data analyzed came from 4 yr of stirred-curd Cheddar cheese production in an industrial cheesemaking facility, comprising over 10,000 vats. Single vat outliers were detected using a simple statistical criterion of mean +/- 3.6 SD on single variable distributions, Fourier series modeling of seasonal variables (fat, protein, lactose, and total solids in milk, and protein in whey), and the multivariate Mahalanobis outlier analysis. Detection of outlier productions (corresponding to several vats) was done by applying the mean +/- 3.6 SD criterion to variables obtained through calculating the fat mass balance, fat retention coefficient, and yield efficiency. Data treatment enabled the detection of outlier data, but also pinpointed variables with a low reliability (manually registered times). Single variable and multivariable methods proved complementary, and the use of both types of methods is recommended when validating an existing database.
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 | 0.138 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.003 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| 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".