Comparison of K-Nearest Neighbor and Multiple Linear Regression for Beauty Product Sales Prediction

Sales predictions constitute a critical component in establishing and expanding a business entity. Accurate sales projections enhance the quality of decisionmaking processes, increase profitability, and improve customer service results. The main objective of this research is to assess the efficacy of K-Nearest Neighbor and Multiple Linear Regression methodologies in predicting beauty product sales. The methodological approach employed in this research is of a quantitative method. The dataset incorporated encompasses variables such as product specifications, pricing, stock, and number of sold. Data preprocessing methodologies are employed to clean the data, handle missing data and detect outliers. The research findings show that both methods can predict product sales. However, the Multiple Linear Regression model has a better advantage than K-Nearest Neighbor. The Multiple Linear Regression method has smaller average values of RMSE, MAE, and MAPE than K-Nearest Neighbor, which are 4.637, 3.990 and 0.077. Meanwhile, the average value of $\mathbf{R}^{\mathbf{2}}$ generated by Multiple Linear Regression is greater than K-Nearest Neighbor, which is $\mathbf{0. 8 2 0}$. This proves that the Multiple Linear Regression method is more suitable for predicting beauty product sales.