Optimizing LSTM Neural Networks for Resource-Constrained Retail Sales Forecasting: A Model Compression Study

TL;DR

This study explores compressing LSTM models for retail sales forecasting, demonstrating that reducing hidden units can improve accuracy and significantly decrease model size, making them more suitable for resource-constrained environments.

Contribution

It presents a systematic approach to LSTM model compression, showing that fewer hidden units can enhance accuracy and reduce size without sacrificing performance.

Findings

Reducing hidden units to 64 maintains accuracy while decreasing size.

Model size was reduced by 73%, from 280KB to 76KB.

Model accuracy improved, with MAPE decreasing from 23.6% to 12.4%.

Abstract

Standard LSTM(Long Short-Term Memory) neural networks provide accurate predictions for sales data in the retail industry, but require a lot of computing power. It can be challenging especially for mid to small retail industries. This paper examines LSTM model compression by gradually reducing the number of hidden units from 128 to 16. We used the Kaggle Store Item Demand Forecasting dataset, which has 913,000 daily sales records from 10 stores and 50 items, to look at the trade-off between model size and how accurate the predictions are. Experiments show that lowering the number of hidden LSTM units to 64 maintains the same level of accuracy while also improving it. The mean absolute percentage error (MAPE) ranges from 23.6% for the full 128-unit model to 12.4% for the 64-unit model. The optimized model is 73% smaller (from 280KB to 76KB) and 47% more accurate. These results show that larger models do not always achieve better results.