Analysis of Indonesia’s Fish Consumption with Regression Method using Go Language

Abstract

The study is made to predict the amount of fish consumption in Indonesia throughout the years 1960 to current year. The amount of fish production and catches will be used as supplementary information to help validate the fish consumption rate. This study is conducted using the Go programming language to prove that even though Go is a general programming language that is rarely being used for data science, it can still be used to perform analytics and machine learning while out-performing other languages that are usually used to do data science like Python and R. There are two primary datasets that are being used in this study, them being the fish captures dataset and the fish consumption dataset. These two datasets will later be parsed and processed to a single file before being fed to the linear regression and decision tree models to achieve the objective of predicting Indonesia’s fish consumption. The Linear Regression model created from our Go Program has predicted a successful model that has a very low R2 score of the predicted regression value vs the true value. Additionally using Go a Decision Tree model has also been created to further strengthen the results of our models given they agree with each other. Both models actually show very high correlation with their final predictions which is 92%. The result of this study solidifies 3 points and that is that Go is a very capable language to be used for data science, linear regression performs better than decision tree in this given scenario that is being used, and finally the fish consumption rate of Indonesia is rising at a much greater rate the world has seen in 1900s.