DV Group, a digital marketing agency, approached us with a request to develop a Data Science project that would forecast the probability of a product from a particular category being purchased in a given time frame in the future.

‍

About DV Group

DV Group specializes in the development and implementation of performance strategies for various major international brands in e-commerce, and brings new tech products to its local market. They are responsible for putting their clients' products on marketplaces. The clients, in turn, track users' actions on the website and then provide the company with collected data showing which users have purchased which products and from what categories.

‍

The approach

To build the predictive model, we decided to use Python as the programming language and Yandex.Cloud as the platform to deploy all the computational processes. Since the system is designed to handle large amounts of data, we chose Kafka as a message broker. It is convenient because it enables modularity and distributes the load, increasing the system's fault tolerance.

‍

The methodology

We implemented the software that learns from the already collected data and then forecasts the probability of a product from a particular category being purchased. Upon getting a user ID, product category, or brand name, the trained model generates a single number. Specifically, the number is the probability of such a purchase in a certain time period.

‍

Results

The development of the predictive model for DV Group was a success. With this model, they can now forecast the probability of a product from a particular category being purchased in a given time frame in the future. This will enable them to provide their clients with more targeted marketing strategies, boosting their sales and revenue.

‍