Supply Chain Intelligence

Supply chain pipelines utilize Time Series forecasting techniques to predict various entities like demand and inventory. However, the nature of these time series data is often nested and traditional modelling techniques often fail to capture its impact.

Hierarchical Time Series Modelling is a novel technique that aids the supply chain intelligence by reducing overall bias in such cases. Not only does this method improve overall metrics and planning mechanisms, implementations of HTS offer both, the long and the short term visibility of the sales.

For example, the monthly sales of a retailer’s products can be grouped on geographical hierarchies like city, state or country. Similarly, the catalogue of products can be further divided into individual products. By using regular time series forecasting methods, information regarding the relationship with these hierarchies may be lost, resulting in inconsistent predictions.

HTS overcomes these shortcomings by maintaining sanity across every hierarchy – like in our example below, the aggregated prediction on a city level is coherent with the forecast of the state, so on and so forth. Businesses and forecasters alike can benefit greatly by exploring HTS modelling in complex supply chain planning projects.

                                                                                               Hierarchical Time Series Modelling