Features of time series forecasting with SPLUNK

In this paper, the methods of analysis and prediction of time series are investigated. The aim of the work is to determine the model that can most accurately predict traffic changes in the short term, as well as obtain a smoothing of the observed curve without losing intermediate points and hiding peak values. To do this, the general principles of regression are considered, as well as the moving average autoregression model and the integrated moving average autoregression model when working with time series are worked out in more detail. These methods are widely used in the analysis of network traffic, monitoring the status of large complexes and objects. It is worth noting that when using a time series, building a trend line or determining seasonality from a complex analytical problem becomes a mathematical formula for describing non-random components of a non-stationary series. As a result of a comparative analysis, it was decided to use the Seasonal Local Level method, since in this case the most optimal criteria for evaluating the model were obtained. However, it is possible to use the LLP5 model in cases where the trend component is more pronounced.