Forecasting The Future With Splunk
Within any business, it is good to predict the future. How many people will buy ice cream when the temperatures hit 20C? How many users will log into our servers at 9am on a Monday morning. Luckily, data science comes to our rescue, and Splunk integrates with two well-defined methods for forecasting: Kalman Filter; and ARIMA:
With ARIMA (AutoRegressive Integrated Moving Average), we get a number of models, and which are stochastic (randomly generated) in their nature. Rather than defining a model, the Kalman filter is an algorithm and can be used to define a probability of the accuracy of the forecast. Kalman filters are often used in guidance systems, such as for navigation and the control of robots, aeroplanes, ships and spacecrafts. The human body can also be modeled with a Kalman filter for the way that our central nervous system controls our movements.
Overall we have a trend which defines the general change in a time series (such as an increase or decrease at a given rate over time). A seasonal element defines fluctuations for time, such as where ice cream sales will be largest in the July, and lowest in December. The logins to a network will also likely to have a seasonal component…
