Estimating Smartphone Price Ranges Using Machine Learning Models

With the contemporary surge in phone technology and new essential functionalities, a variety of models with varying features and price points are continuously released every year. The exchange market for older devices is also increasingly vibrant. Estimating smartphone price ranges becomes an important task for both sellers and buyers. Minimizing prediction errors or offering a strategic price, helps customers make decisions easier, thereby increasing conversion rates for sellers. However, smartphones exhibit a wide spectrum of vendors, features, specifications, and price points, necessitating effective price is challenging. Machine Learning (ML) models typically perform better than statistical and econometric models [1]. In this research, we compare the effectiveness of five ML models (Ridge Regression (RR), K-Neighbors Regression (KNR), Decision Tree Regression (DTR), Lasso Regression (LR), Random Forest Regression (RFR)). A dataset of about 3000 records with 16 features including brand, price, year of manufacture, ram, battery, etc. are manually collected from Amazon and used for this study. Comparative results are analyzed concerning the highest achieved r2_score. The findings show a strong correlation between smartphone prices and factors such as brand, operating system, battery capacity, memory, screen size, camera features, color, weight, and warranty period. And the Lasso Regression outperformed the other models with an r2 score of 0.97584.