SMART Demand Forecast – a demand forecasting system based on machine learning and artificial intelligence algorithms. Release 1.0

Inventory is one of the most valuable assets of a company. The key to proper management of which is high-precision demand forecasting. SMART Demand Forecast, based on machine learning algorithms and artificial intelligence, helps Supply Chain Managers with this. The solution allows you to determine the optimal quantity of goods to be ordered, sufficient to meet demand, and prevents overstocking in the warehouse. All this, in turn, affects proper availability in retail outlets, reducing the level of lost sales and at the same time reducing the level of funds frozen in inventory.

The relevance and necessity of a demand forecasting system is beyond doubt, so we invite you to look into the functionality of SMART Demand Forecast.

SMART Demand Forecast functional blocks

Operation of SMART Demand Forecast is ensured by 5 blocks: modeling, selection of analogs, analytical reports, processing of personal information and system configuration.

Information processing block

The system operates on the basis of machine learning and artificial intelligence algorithms, for which the input data for model training plays an important role. Therefore, we paid special attention to verifying their integrity and the possibility of secure processing, as well as storing historical data of business activities and personal information of users. Let’s take a closer look at the features of this block:

  1. Processing of personal data in accordance with the GDPR

SMART Demand Forecast takes care of security of received data and complies with the General Data Protection Regulation, which is why:

  • Authorization in the system occurs using Microsoft Azure Active Directory.
  • Personal information about users, which is processed by the system, is pseudonymized and stored directly in Microsoft Azure Active Directory.

  1. Data integrity check

The system incorporates intelligent algorithms for processing large arrays, which makes it possible to check integrity of data. This allows you to detect problems with the data even before running the model.

Modeling block

SMART Demand Forecast has a modeling block that allows you to start the process of training the model and preparing a forecast, administer promotional campaigns, and save and export the results. The block has the following capabilities:

  1. Filtering by products and stores

In accordance with the levels of the product hierarchy and the levels of filtering by stores, the system provides for the use of filters in the modeling block for the information displayed on it.

  1. Starting the model training process

In the demand forecasting system interface, users with the appropriate access rights can start the model training process.

  1. Displaying the stages of model training and forecast preparation

In SMART Demand Forecast, end-to-end display of the model training and forecast preparation statuses are implemented for all users.

  1. Cancelling the process of training the model and preparing the forecast

The ability to stop the model training and forecast preparation processes if necessary is implemented for users with appropriate access rights.

  1. Import of promotional campaigns

The interface has a feature of importing promotional campaigns using a pre-exported template file.

  1. Forecast preparation and its further saving

Users with the appropriate role rights can:

  • Start the process of preparing forecasting using machine learning algorithms,
  • Finalize the prepared forecast,
  • Export results in CSV format.

Analog selection block

In SMART Demand Forecast, functionality for analog processing is implemented. Users have the ability to:

  • Check products / stores, that do not have enough data on actual sales to forecast, for availability,
  • Analyze the list of proposed anaogs of products / stores,
  • Select a relevant product / store as a benchmark.

Analytical reports block

Analytics is an integral part of the system. Reporting in the system is based on Power BI, integrated into the interface, providing flexible and complete information. Users can analyze:

  • Historical sales,
  • Forecast quality based on historical data,
  • Sets of promotional campaigns involved in forecasting,
  • Compensated sales.

System configuration block:

Custom settings of SMART Demand Forecast make it more flexible for your business. You can:

  • Make settings related to training and forecast preparation,
  • Monitor system users, their rights and statuses,
  • Select personal settings,
  • Set your own filters with different sets of conditions for further use in the modeling block,
  • Set up planning options, including:
    • Opening date of the planning period,
    • Planning horizon,
    • Required time series for model training,
    • Forecast granulation level

Want to learn more about SMART Demand Forecast?

Email us at and we will contact you.

Posted in: