Novis is leader in the implementation and management of SAP solutions and digital innovation.
Currently, they have over 150 clients in Chile and Latin America, managing more than 300 instances and 30.000 users, with 150 consultants and SAP managers staff.
Several Novis’ customers face the challenge of obtaining reliable sales forecasts for multiple locations on a large set of different products.
Our solution consists in the implementation of a recurrent neuronal network algorithm (DeepAR) using Sagemaker.
The output corresponds to a probabilistic distribution of predictions for each one of the forecasted time series (unique combination of products and locations), which facilitates the management and optimization of the relationship between stock-break and over-stock, without the need to incorporate a greater complexity when making decisions, nor additional human resources.
The solution enables Novis to offer to their clients a more precise and faster of implement forecasting tool, thanks to the benefits of SageMaker including:
Fully managed training and hosting
Easy to scale
75% lower inference costs
- Decreased time spent on forecast preparation.
- Improve business planning.