| Title |
Applying ensemble clustering methods to a complete blood count test results using openensembles: a python resource for ensemble clustering / |
| Translation of Title |
Ansamblio klasterizavimo metodų taikymas bendrojo kraujo tyrimo rezultatams, naudojant „Python“ biblioteką „OpenEnsembles”. |
| Authors |
Petrišiūnaitė, Gintarė |
| Full Text |
|
| Pages |
36 |
| Keywords [eng] |
Ansamblio klasterizavimo metodai, Python, OpenEnsembles, ensemble clustering |
| Abstract [eng] |
Applying Ensemble Clustering Methods to a Complete Blood Count Test Results Using OpenEnsembles: A Python Resource for Ensemble Clustering Clustering algorithms have their individual limitations. In order to overcome limitations of individual clustering algorithms and improve clustering results ensemble clustering method can be applied. Ensemble clustering method combines results of cluster partitions of different clustering algorithms and can produce the final clustering solution of an improved quality. This thesis focuses on applying ensemble clustering methods to a Complete Blood Count test parameters. Reference intervals were calculated and used for cluster validation after outlier removal, using Tukey fences. Ensembles were created using a Python library OpenEnsembles. Density based clustering algorithms were able to detect outlier clusters. Ensembles created using co-occurrence linkage ensemble creation method avoided over-merging or under-merging clusters unlike majority vote and graph closure ensemble creation methods. |
| Dissertation Institution |
Vilniaus universitetas. |
| Type |
Master thesis |
| Language |
English |
| Publication date |
2020 |
