Matthew Guevara
LSAMP Scholar
Major: Computer Science
Mentor: Roni Khardon
The purpose of this study is to learn machine learning algorithms conceptually and in practice. This involved applying these methods to various datasets from the recent paper DeepMicro: Deep representation learning for disease prediction based on microbiome data. I delved into a specific type of machine learning known as supervised machine learning and used this technique to assess microbiome data. In order to accomplish this, the study utilizes scikit machine learning algorithms within python. The study found that the performance of these basic algorithms was consistently lower than the more sophisticated approach of DeepMicro. This shows the value of that method and the need to tune setting of machine learning algorithms, which we did not perform in this study. Among the simple methods RF classifiers performed better than other algorithms, but the variance in multiple runs with cross validation is high so that differences may not be significant.
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. IN LSAMP is supported by NSF Award Number: HRD-1618408, 2016-2022.
Leave a Reply