One of the largest apparel retailers in the US had been experiencing uneven growth across its geographic markets. Some markets were showing rapid growth with minimal marketing efforts while others were not showing similar trends in spite of focused marketing efforts. The client approached AbsolutData to help them identify the reason(s) behind this.

• The project was delivered to the client in 6 weeks, with the majority of the time being spent in getting client data ready for modeling.
• The team at AbsolutData analyzed the data of each region and soon realized that different regions had different reasons/ factors that influenced the market growth which resulted in irregular growth across regions. The sales prediction model being used by the client was built without incorporating differences of regions and was accurate in only 23% of the regions.
• In order to attain high accuracy, different regions showing similar characteristics were grouped into different segments. AbsolutData built a sales prediction model which took such differences into consideration by using a cutting edge technique called Latent Class Regression. This involved fitting separate regression equations to the different segments. This pointed out that drivers are different across the different segments. Using this technique AbsolutData created prediction models for different markets, resulting in an accuracy level of over 70%.

The project deliverables to the client included:

• Model Parameters, Model Description and Model Accuracy measurements and recommendations on Model implementation.
• Model Performance Matrix that tracks model prediction versus actual behaviour.
• Management Dashboard Report that highlights key metrics.