ARMDN: Associative and Recurrent Mixture Density Networks for eRetail Demand Forecasting
This work addresses demand forecasting for online retail organizations to improve supply-chain planning, representing an incremental advancement by combining existing techniques like embeddings, MLP, LSTM, and mixture density networks.
The paper tackles the challenge of forecasting demand in online retail, where traditional methods fail due to associative factors and non-stationary shifts, by proposing AR-MDN, a neural network that models these factors, time-series trends, and variance, resulting in significant accuracy improvements on a dataset from Flipkart.
Accurate demand forecasts can help on-line retail organizations better plan their supply-chain processes. The challenge, however, is the large number of associative factors that result in large, non-stationary shifts in demand, which traditional time series and regression approaches fail to model. In this paper, we propose a Neural Network architecture called AR-MDN, that simultaneously models associative factors, time-series trends and the variance in the demand. We first identify several causal features and use a combination of feature embeddings, MLP and LSTM to represent them. We then model the output density as a learned mixture of Gaussian distributions. The AR-MDN can be trained end-to-end without the need for additional supervision. We experiment on a dataset of an year's worth of data over tens-of-thousands of products from Flipkart. The proposed architecture yields a significant improvement in forecasting accuracy when compared with existing alternatives.