The BI platform would use open-source data from online stores and marketplaces about the popularity of certain styles in places of sales interest to generate predictions for designers, merchandisers, and buyers.
The predictions would allow fashion brands to adjust production and sales strategies based on the knowledge about which items and styles are likely to be in high demand next season.
Validating product concept
●
ITRex’s business analyst kicked the discovery stage off by investigating the sources of data for powering the business intelligence platform. The research centered around open-access data from online sales platforms and marketplaces, for example, Shopify, Farfetch, etc. The data would give insight about products, sales, and factors that influence sales.
●
The business analyst working together with an ML engineer and a solution architect nailed down the logic behind the ML model. Based on the researched data, they could extrapolate demand for certain products or styles at certain places of sales interest during a particular season for a particular customer category. The extrapolation logic was proven a success following several tests.
●
As a result of the activities above, we confirmed the viability of the customer’s vision, proving that there is enough open-access data to build a predictive model with the algorithm of choice.
Crafting functional solution
●
The discovery team described and visualized the BI platform's functional blocks, including back office, billing, reporting, legal and compliance, and others. They also identified functional priorities for developing a minimum viable product and a minimum valuable product.
●
We followed by nailing down an exhaustive functional requirements document.
●
As the BI platform relies on complex data flows and comprises several functional blocks, we designed a flexible architecture. The architecture boasted built-in flexibility that allowed expanding the number of data sources to power the prediction engine.
●
As the customer relied on open-source data and was unfamiliar with data compliance requirements, we put together data collection and storage recommendations to make sure the data used for machine learning was handled securely.
●
We finished the discovery stage by crafting a thorough product testing strategy.
