Advanced AI for Demand Forecast, Assortment Planning and Plan Monitoring in Fashion and Apparel Retailing

Retailers require reliable demand forecasts for their operations management and planning. Demand forecasting for fashion products is, however, an extremely challenging task. A good solution for this problem should address at least the following three questions: (i) high volatility of demand and its dependence on external factors (ii) forecasting flexibility for different spatio-tempo-hierarchical aggregation levels, and (iii) forecasting for new products without historical data. The other aspect of the problem is to follow the gradual actualization of the demand in time, update the forecasts, and detect anomalies. The outliers and anomalies could subsequently be translated to corrective responses to meet the demand and minimize costs and lost opportunities. The main objective of the project is to explore the state-of-the-art algorithms for time-series forecast and anomaly detection followed by the design and implementation of a demand forecast and monitoring framework that could appropriately address the three above-mentioned forecast challenges and detect anomalies in demand.

Faculty Supervisor:

Nizar Bouguila

Student:

Ornela Bregu;Rafiul Hasan Khan

Partner:

FIND Innovation Labs Inc.

Discipline:

Engineering

Sector:

Professional, scientific and technical services

University:

Concordia University