Before starting with the model application, let us take a look at how those durations behave in time scale and how they depend on previous one. In fact we can consider few plots to estimate as follows:
Based on these two charts we will estimate and model a behavior of the durations.
The Autocorrelation function is often used in signal processing and is a correlation of time series against itself lagged by a few steps. Usually for discrete observations the estimate of autocorrelation function is obtained as follows:
In the preceding equation, is a lag ...