This paper aims to analyze the differential purchase intentions of consumers in an e-commerce context. This is inspired by the works of recent studies that consider factors like unit ‘price’, product ‘review’, and attributes of the product image in websites, including the ‘Brand’, ‘Background’ image, ‘Promotion and Advertising’ content, and presence of mannequins/ ‘model’. Existing studies are found to analyze consumers' purchase patterns but do not predict the product's demand. The demand needs to be estimated to make more informed marketing decisions regarding product design and development. The demand is now predicted (number of transactions) as ‘Deal’ for male and female consumers using advanced machine learning (ML) algorithms like random forest, gradient boosting, support vector machines, and deep neural networks. Though existing studies are found to compute variable significance using a hierarchical regression model, the significance must be validated mathematically, statistically, and from a stakeholder's perspective in uncertain scenarios. The paper overcomes this limitation by computing the variable importance from ML and further statistically validating through multiple linear regression. The study findings provide valuable insights for both customers and website product designers.