Emerging Trends Demand Forecast using Dynamic Time Warping

Forecasting demand for trends in their introduction phase plays an important role in strategic planning for product manufacturers. Lack of sufficient historical data in the early stages of a trend makes it difficult for standard time series and statistical methods to do a robust forecast. In this paper, we have proposed a novel approach to forecast the demand of emerging trends. Our methodology involved using Dynamic Time Warping (DTW), a pattern matching algorithm to identify similar trends for historical matching time frames, and used as leading indicators in a regression model to forecast demand. The forecast accuracy of these models was found to be significantly better when compared with popular forecasting methods like Autoregressive Integrated Moving Average (ARIMA), Prophet, and Long Short-Term Memory (LSTM). We have taken the Food and Beverage industry as an example application area to explain our methodology.