Solutions for Real-Time Locating Systems


Use Cases

by infsoft


Machine Learning in Shopping Centers

Applying machine learning to shopping centers allows a precise forecast of the number of visitors that can be expected. On that basis, advertising spaces can be priced accurately.

At a glance
  • Application of machine learning in shopping centers
  • Correlation of weather data and number of visitors
  • Automatic determination of prices for advertising spaces

Problem Definition

The number of visitors in shopping centers varies all the time. At times when there are a lot of visitors, advertisements, for example on advertising displays, are more likely to be successful as they are seen by more people. However, the operators of shopping centers are not able to predict the number of visitors very well. As a result, an adequate and fair pricing of advertising spaces can’t be achieved.


Implementing a machine learning tool solves this problem. Machine learning enables a reliable forecast of the number of visitors and thus appropriate pricing of advertising spaces.

A factor that affects the number of visitors significantly is the weather. When the weather is bad and it’s raining, there are usually more people visiting the shopping center. However, with this information alone the human brain can forecast the number of visitors very roughly at best. Machine learning on the other hand recognizes complete and precise correlation between the weather and the number of visitors and can make very accurate forecasts on that basis. With the knowledge of current weather data, the system can predict the number of visitors at a certain time accurately. Subsequently, adequate prices for advertising spaces can be determined automatically.

Technical Implementation

Machine Learning in Shopping Centers

The available data sets are essential for any machine learning system. The quality and quantity of data determines the accuracy of the machine learning results. In this use case, the number of visitors can be measured using indoor positioning with Wi-Fi while the weather-related data can be collected using infsoft Sensor Tags.

The data sets are imported into the infsoft Machine Learning Tool and given to an algorithm. With the historic data, the system gains experience and is trained. The gathered experience enables the model to make precise forecasts of how many people will visit the shopping center when it is given input of current weather data. The predicted number of visitors is then provided to infsoft Analytics and the Automation Engine. It is previously defined in the Automation Engine, which price for advertising spaces should be set when the forecasted number of visitors is located in a certain range. It is then possible to send an automated e-mail containing the price for an advertising space to the responsible persons and the shop operators.

Thanks to the increasing amount of weather and visitor data, machine learning can gather more experience and forecasts will get even more precise over time. Beside the weather other factors like school holidays or big events in the area may affect the number of visitors in a shopping center. Those data sets can also be defined in the system to obtain even better predictions.

Use Case PDF for Download

Get the white paper for free

Great introduction to the topic of indoor positioning

Comments are closed.