Forecasting SKU Demands in the Manufacturing process

Our Client

Multi-store retailer with multi-product manufacturing capabilities

Problem Statement

  • A Multi-Store Retailer experienced unanticipated volatility for the demand of a few specific SKUs from a single product category.
  • The client aimed to create a forecasting model to predict demand for certain SKUs for the next 1 to 12 months. Monthly data for the price, sales, and around 50 current-period or lagged potential predictor factors were included in the time-series dataset.
  • Rapid middle-class wealth growth in India produced periodic surplus demand and giant surges for certain SKUs in the client’s manufacturing process.
  • Furthermore, fresh supplies of identical items in the market appeared often, resulting in periodic overstock and price reductions due to competition.

Our Solution

The steps that we followed to execute our solution were as follows:
  • An ensemble of LSTM and Autoregressive Time-Series Model was built for forecasting future demand. The client business used forecasted demand to have greater control over the production and inventory costs, and increase profitability.
  • Our client utilized it to forecast future demands to manage the product stocks and increase their profitability.
Analytical Planning
  • The Analytics team created a time-series analysis dataset, converting all series to monthly data and accounting for missing values, holidays, and sales seasons, among other things.
  • The most promising linear, non-linear predictors, lagged predictors, and combinations of predictors were identified using variable selection methods on more than 20 macroeconomic variables. The predictive power was calculated based on the mean absolute prediction error.
  • A forecast simulator based on an ensemble of LSTM and Autoregressive Time-Series Models got molded to forecast 1, 2, 3, 6, 9, and 12 months into the future while accounting for serial correlation (the correlation over time of the impact of unobserved variables on the data being predicted—in this case, demand).
The client’s management team was able to enter updated values of predictor factors each month and anticipate the future demand of the specific SKU using a Web-based forecasting tool. The client organization also evaluated the model by comparing anticipated and actual values for the first several months. The customer firm was convinced with the prediction accuracy that they:-
  • Incorporated model projections into company processes.
  • Carried out follow-up research using the forecasting approach on a different product category.
  • It contributed to their revenue increase of about 12-14%.

Facing a similar challenge in your business?

Business Impact

The objective was to create a demand forecasting model for a specific product category SKUs.
  • Collecting and compiling a database of micro, macroeconomic time-series indicators, and transactional data were used to graph the anticipated demand.
  • Use time-series regression and deep learning tactics to create a reliable demand forecasting model for a given SKU.
  • Create a web-based forecast simulation tool that allows clients to input updated predictor factors and evaluate updated SKU demand projections.

Looking for a Similar Solution?
We Can Help Protection Status