K-Nearest Neighbors Algorithm for Prediction Shelf Life of Rice Based on Electronic Nose Dataset

Abstract

In Indonesia, rice is a food commodity that has a strategic and vital role. Considering rice's importance, the government always strives to ensure food needs and a surplus of rice as food reserves. However, rice has decreased in quality and is not suitable for consumption in recent years. Conventionally, the rice shelf life prediction methods use the direct method that the rice samples are tested by smelling the rice using the human sense of smell to predict how long rice's shelf life is. Therefore, we propose another method to predict how long rice's shelf life. Developing a prediction system for the shelf life of rice uses the k-nearest neighbors (k-NN) algorithm and electronic nose (E-nose) dataset to predict how long rice's shelf life more quickly. This experiment showed that the k-NN Regression algorithm obtained the best parameters with the R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> score of 0.7217 and the RMSE score of 3.8043. This method predicts the shelf life of rice effectively and solves existing problems because it can achieve accuracy very well.