Can someone share some reference materials that explains the graphs of autocorrelation used in the lecture? Particularly what is the reason for spikes and decay when using autocorrelation. What are visual clues to know whether the graphs is the result of autocorrelation?

Please see the ungraded labs for the week 1 and reply if you need more help.

Hi @balaji.ambresh yes I already did the ungraded labs. My question is about building intuition with graphs especially the reason for spikes and decay when using autocorrelation, the visual clues to look for whether the graphs is the result of autocorrelation or is it just seasonality?

I don’t think you figure out the lag value of correlation just by looking at the graph.

That said, consider the following method:

- Let’s say that you expect the current data point to be computed by some function of 50 steps before.
- Create a pandas dataframe with lagged values by using
`shift`

function for lags in range [1, 50]. - Invoke
`dataframe.corr()`

to produce the correlation information. - The non-diagonal entry with the highest value is the best guess for the lag.