Non-seasonal differencing for stationarity

Differencing is a way of making a time series stationary; this means that you remove any systematic patterns such as trend and seasonality from the data. A white noise series is considered a special case of a stationary time series.

With non-seasonal data, you use lag-1 differences to model changes between observations rather than the observations directly. You have done this before by using the diff() function.

In this exercise, you will use the pre-loaded wmurders data, which contains the annual female murder rate in the US from 1950-2004.

Plot the wmurders data and observe how it has changed over time.
Now, plot the annual changes in the murder rate using the function mentioned above and observe that these are much more stable.
Finally, plot the ACF of the changes in murder rate using a function that you learned in the first chapter.

Exercise

Non-seasonal differencing for stationarity

Instructions