Analysis of K-NN Algorithm and Linear Regression to Predict House Prices in Jabodetabek

Nadia Putri Ariyanti (Faculty of Communication and Information Technology, Universitas Nasional, Jakarta, Indonesia)
Agung Triayudi (Faculty of Communication and Information Technology, Universitas Nasional, Jakarta, Indonesia)
Ratih Titi Komala Sari (Faculty of Communication and Information Technology, Universitas Nasional, Jakarta, Indonesia)

Journal of Blockchain, Nfts and Metaverse Technology
Vol. 2 No. 1 (2024): February 2024

Jabodetabek is now the region with the highest average level of citizen satisfaction, so many people migrate to this region in the hope of getting better living conditions, this will make people who want to buy a house question whether the house they want to buy is good value or not. The purpose of this study is to evaluate the effectiveness of multiple linear regression and K-Nearest Neighbors (KNN) algorithm on a dataset of house prices in Jabodetabek. Better results are obtained by using the Multiple Linear Regression model which has lower Mean Absolute Error (MAE) and Mean Squared Error (MSE) values and a fairly good R-squared of around 48.72%. However, the very high MAE and MSE values of the KNN model indicate inaccuracy and significant prediction variance. Although KNN has a relatively high R-squared value, more research is needed to see if the model can adequately explain data fluctuations. Based on the performance evaluation, multiple linear regression is ultimately a better choice