Problem Statement Since the Covid-19 pandemic, consumer behavior in purchasing attire has drastically shifted from traditional retail settings to eCommerce. Retail companies need to adapt to the new norm and create a better online shopping experience for shoppers. The product recommendation system,which suggests the most suitable products to consumers based on past purchases and other algorithms, is one of the most critical functions in eCommerce. However, the first step to better serve the customers is to understand them on a deeper level by utilizing data to generate insights. According to Retailweek, H&M is the world’s second-largest clothing retailer. As an industry leader, it is crucial to be at the forefront of the latest technologies and provide the best online shopping experience. Consequently, we propose constructing an organized database for the client and product data at H&M to help the company create a more effortless and accessible understanding of their customers.
Proposal We decided to utilize the H&M Personalized Fashion Recommendation datasets provided by the H&M Group on Kaggle. We would like to use the information provided by these datasets to analyze customer preferences and ultimately predict future purchases based on the data from previous purchases. From our research, we would be able to understand our customers’ purchasing decisions more profoundly over time and therefore provide better products in the future. Companies can utilize the data collected from their customers to generate insights into the interests and lifestyles of the different customer segments and therefore forecast future sales. Overall, this insight informs future business decisions and strategies, making our analyses' results in this project pertinent to H&M. More specifically, the analysis will help H&M understand if their current efforts align with customer demands. Do these companies offer the kinds of products their customers are looking for? Are these companies able to keep up with trends and changing needs? The analysis could also support other departments such as the marketing team to understand what products they should advertise to different customer segments or the IT department to improve online recommendations. To meet our goals, we will begin by extracting valuable attributes and combining them into unique and practical tables that satisfy normalization rules. These tables would connect to the products, consumers, or transactions, or they will be a relationship table that explicitly connects the tables. After creating our tables, we will find trends or similarities across the different types of products and consumers by using Python. Then we will use the programs to connect the products and consumers to transactions to understand consumer purchasing decisions.