Neural Networks

An artificial neural network is a mathematical model, as well as its software or hardware implementation, built based on the organization and functioning of biological neural networks—networks of nerve cells in living organisms. This concept originated from studying processes in the brain and attempting to model these processes.

Neural networks in MySales are used both for building multifactor sales forecasts for items where it is important to determine the synergy effect of two or more factors, such as seasonality and discount, and for forecasting promo uplift for promo items where no comparable or similar past promos can be found.

The key advantage of neural networks over regression analysis methods in sales forecasting is their ability to identify synergy effects. However, this same advantage leads to the main drawbacks of neural networks: higher susceptibility to overfitting, low transparency, and limited interpretability of the results produced by the network.

The proprietary neural network algorithms in MySales are distinguished by their ability to train the network within predefined rules, making the results more understandable to users and predictable.

Additionally, MySales allows visualization of the neural network trained for promo uplift forecasting, showing which factors are included as input neurons and how they influence the final output (two output neurons). This helps make the neural network training results more interpretable and clear to the user:

Open