Abstract-Data mining tools are nowadays becoming more and more popular in the semiconductor manufacturing industry, and especially in yield-oriented enhancement techniques. This is because conventional approaches fail to extract hidden relationships between numerous complex process control parameters. In order to highlight correlations between such parameters, we propose in this paper a complete knowledge discovery in databases (KDD) model. The mining heart of the model uses a new method derived from association rules programming, and is based on two concepts: decision correlation rules and contingency vectors.

Information on partial lactations often is included in genetic evaluations by pre dicting the cow's eventual 305d yield. Such projected yields have less phenotypic and genetic variation than completed yields but were modeled as having greater or equal variation in evaluations. Analysis of first lactations from 48,424 daughters of 844 Holstein sires indicated that yields predicted early (46 to 75 d) in lactation had less than one-half as much additive genetic variance as completed yields. Multiple-trait REML estimates of genetic correlations of predicted and completed yields were all above .92, indicating that early lactation information is valuable if modeled appropriately. Expanded records with genetic variances equal to those of completed yields and new lactation length weights were derived. Expanded records have larger error variances than either projected or completed yields and, thus, are given less weight when included in animal model evaluations. Genetic gains are expected to increase only .2 to .3%, but more stable genetic evaluations should result from use of expanded records, particularly for animals evaluated primarily from first lactation records in progress.

Lactational means of somatic cell score from sample days of primiparous DHI cows were analyzed to compare dairy records processing centers for data properties, to examine effects of age at calving and month of calving on somatic cell score, and to calculate preliminary PTA of sires for somatic cell scores. Five processing centers contributed data but differed substantially with respect to availability of records in progress, DIM on last sample day, and length of time that data were submitted. Highest lactational means of somatic cell score tended to be in short lactations for young cows but in long lactations for older cows. Regions of the country differed substantially for effect of month of calving on lactational means of somatic cell score, but lactational means of somatic cell scores increased with age at calving for all dairy records processing centers. Standardized yields of milk, fat, and protein were included as multiple traits with somatic cell score for analysis. Heritability of somatic cell score was .08 to .16 for individual processing centers and .10 across processing centers. Phenotypic correlations of somatic cell score with yield traits were negative, but genetic correlations were positive. Selection for decreased somatic cell score on a national basis should be possible and would seem advisable to decrease mastitis infections. Response may be slow, however, because of antagonistic genetic relationships with the yield traits, which have great economic importance.

No abstract