Predicting Laptop Prices Using Machine Learning and Feature Engineering

The prices of laptops are highly diverse depending on a blend of the hardware requirements and the brand position, and as such both consumers and retailers find it hard to ascertain their actual worth and the retailer to price the products competitively. This paper uses the machine learning method to estimate laptop prices based on features of the product in detail, using the publicly available Brand Laptops Dataset available on Kaggle, which includes specifications of 991 laptop models. The information contained on the dataset will be processor setup, memory size, storage format, monitor size, operating system and brand name. Following the initial data cleaning, the heavy feature engineering was carried out in the presence of the additional informative variables like the numerical indicators of CPU tier and the operating system type. Unnecessary columns were eliminated, and all the numerical variables were standardized to counterbalance the same contribution as they contribute to the training of the model. This was followed by the development of a linear regression model that was used to estimate price based on these engineered features. An 80/ 2 0 train- test split was used to assess model performance. The model on the training set got a Mean Absolute Error (MAE) of around 17,261, and the validation set got 15,702, which represents an average prediction error of 20 percent of the cost of the laptop. These results indicate that a fairly straightforward regression expression is capable of creating key pricing patterns motivated by fundamental elements of performance. The suggested solution offers a benchmarking tool that could help both consumers and retailers in determining the price fairness and/or the pricing strategy optimization. The future enhancements might consist of more sophisticated machine learning models and other functions like the GPU classification and the build quality metrics.