Automated feature selection for a machine learning approach toward modeling a mosquito distribution

- A data science technique is used to find an optimal set of inputs for SVM training.
- The input selection of the data science and of the biological experts differed.
- The data science approach could help to find new knowledge.
- The pros and cons of the used data science approach are discussed.
- The implementation was made on a cluster computer based on open source software.

This paper introduces a data science method to determine a set of features for training a vector support machine (SVM). The SVM is used to model the relationship between the distribution of one particular invasive mosquito species and climate data. Two biologists selected training data on the basis of their domain expertise. This was compared with the result of the data science simulation. The paper then explores the possible uses of data science to generate new knowledge as well as to identify the weaknesses of this technique.