Documentation

Pattern Recognition Alert

This machine-learning algorithm applies pattern recognition and forecasting techniques to predict the next expected value of a data series. The underlying data set thereby needs to show a clear pattern and trends to make accurate predictions about the behavior. Generally, the stronger the pattern the more accurate the predictions.

 

Using this method, we will predict the value of the next period based on your historical data and then compare this expected value to the real outcome. A significant deviation between the actual value and the expected number, may indicate an unexpected behavior as it does not fit into the known pattern. You may adjust the sensitivity of the algorithm by defining the range of the allowed deviation from the forecasted value and thereby adjust the number of alerts you will receive.