Preprocessing and Machine Learning classification for targets extracted from Multi Beam Echo Sounder data

This code was written for the purpose of classify through Machine Learning techniques data coming from Multibeam Echosounder surveys in order to automatically identify gas, fish schools, platform and noise. 
The notebook_19032021.py file is exported from a Jupyter notebook used to perform preprocessing, feature engineering, exploratory data analysis and predictions. The dependency from Orange (Conda suite) present in the previous version of the code has been dropped in order to make the tool more flexible and customisable.

The input data is released under CC-BY 4.0 license and it is available here: https://doi.org/10.17882/79142