Comparison of Hybrid Novel Pearson Correlation Coefficient (HNPCC) with K-Nearest Neighbor (KNN) Model to Improve Accuracy for Movie Recommendation System

Abstract

A hybrid recommendation model based on the HNPCC and the K-Nearest Neighbor (KNN) model were evaluated to increase movie recommendation accuracy. The information gathered from the movielens dataset, which contains 23 attributes and with 30 samples, for use in a hybrid movie recommendation system. The sample size for each set is 30 people, and pre-test power is 0.8.Using an independent t-test to decide statistical significance with p<0.05, it was found that HNPCC has a slightly higher accuracy of 94.3% significantly, while KNN has a lower accuracy of 92.9%.As a result of the comparison, the HNPCC outperforms the KNN in terms of enhanced accuracy.