Product pricing solutions using hybrid machine learning algorithm

Area of Science:

• E-commerce analytics
• Machine learning applications
• Computational economics

Background:

• E-commerce has grown significantly, especially post-COVID-19, increasing the complexity of product pricing.
• Traditional pricing strategies struggle with vast product numbers, seasonal trends (apparel), and fluctuating specifications (electronics).

Purpose of the Study:

• To develop a competitive product pricing strategy for e-commerce businesses.
• To leverage product reviews, statistical, and categorical features for accurate price prediction.
• To introduce and evaluate a novel hybrid machine learning algorithm for e-commerce pricing.

Main Methods:

• A hybrid algorithm, X-NGBoost, combining extreme gradient boost (XGBoost) and natural gradient boost (NGBoost), was developed.
• The model utilizes product reviews, statistical, and categorical data for price prediction.
• Performance was benchmarked against ensemble models: XGBoost, LightGBM, and CatBoost.

Main Results:

• The proposed X-NGBoost model demonstrated superior performance compared to existing ensemble boosting algorithms.
• The hybrid approach effectively captures complex pricing dynamics influenced by product attributes and customer feedback.
• Accurate price prediction was achieved, aiding competitive market positioning.

Conclusions:

• X-NGBoost offers a robust solution for dynamic e-commerce product pricing.
• The model provides valuable insights for businesses aiming to optimize pricing strategies in competitive online markets.
• Leveraging machine learning enhances pricing accuracy and competitiveness in the e-commerce landscape.