Using machine learning on a diverse class of problems : from rainfall to criminal actions

Abstract

We intend to compare and analyze certain machine learning algorithms by taking two different datasets tackling separate real world issues. The first one relates to agriculture. Bangladesh, being an agrarian country, is heavily dependent on rain. Being able to predict the rainfall amount accurately would enable successful and sustainable production. Machine learning algorithms can also help predict the category of crimes in a particular area. This will enable law enforcers in a certain region to predict and categorize recent crimes based on past incidents. Our aim was to approach these problems by labelling and processing the data and then comparing the results of different cost functions.