Autogating in Flow Cytometry Data using SVM Classifiers for Bacterioplankton Identification
Flow Cytometry. Support Vector Machine. Enviromental Analysis. Autogating. Machine Learning.
This master tesis shows the results of a methodology proposal for bacterioplankton identification using a machine learning approach named SVM. Samples used were taken from 19 high elevated lakes located at Pyrenees Mountains. Samples generated 74 databases after been analyzed by a specialist to serve as input to the algorithm. We observed the viability of this method with 3.35% of error in identification. Furthermore, there is no isolated direct correlation between robustness of the prediction models and high complexity of the input data but, indeed, the algorithm settings, function cost and variables choice have an important role in the performance as well.