Modular framework for outlier detection

Outlier detection is an important form of data analysis because outliers in several cases contain the interesting and important pieces of information. In the recent years, many different outlier detection algorithms have been devised for finding different kinds of outliers in varying contexts and environments. Some effort has been put to study how to effectively combine different outlier detection methods. The combination of outlier detection algorithms as an ensemble was studied in this thesis by designing a modular framework for outlier detection, which combines arbitrary outlier detection techniques. This work resulted in an example implementation of the framework. Outlier detection capability of the ensemble method was validated using datasets and methods found in outlier detection research. The framework achieved better results than the individual outlier algorithms. Future research includes how to handle large datasets effectively and the possibilities for real-time outlier monitoring.

Detection of bovine foetal DNA from amniotic fluid using the polymerase chain reaction

Selostus: Sikiön DNA:n tunnistaminen naudan sikiövedestä polymeraasiketjureaktion avulla

Detection of bias in animal model pedigree indices of heifers

Selostus: Eläinmalliin perustuvien hiehojen odotusarvojen luotettavuus jalostusarvon ennusteena