Energy Logserver version 7.4.2is equipped with the Empowered AI module https://energylogserver.com/empoweredai/. This is the result of many months of work by a team of Mathematicians, Data Scientists, Architects and Programmers.The work is financed from the EU grant awarded to Energy Logserver by the NCBiR unit. Empowered AI is an engine of mathematical algorithms whose goal is to discover knowledge based on collected logs. Update 7.4.2 provides the first data analysis scenario, which is the forecast.
In Empowered AI – Forecast, numerical values that we can predict are all numbers in our logs. These may be bytes transferred, values based on netflow data, the number of application sessions or user statistics. A special numerical value is the number of selected events – the so-called the doc_count. Let’s look at the time distribution of events coming from e.g. a mail server.
Let’s analyze only the doc_count numerical distribution for this data:
We see increases in the number of events that tell the story of a typical work day. This course shows us the correct state that the postal operator expects every day of his work. This distribution can be called a behavior pattern. A DevOps knows how source systems behave because he accumulates the experience. Let E-AI do the same. By running Forecast analysis, we collect experience. We build the model of this behavior as a model course of one day. Here is the result of the working day forecast:
Empowered AI is based on source data and the forecast records values in the future. We are dealing with generated samples, which are further processed in the same way as the original data. Therefore, by using the visualization of differences, we can easily refer the forecast model values to the data that will arrive in real time. This will be the moment to check both the quality of the predictions and also to evaluate the incoming data whether they are consistent with the behavior pattern or deviate from it.
The behavior example for the entire week :
