0
$\begingroup$

I have two time series I want to prove if they are stationary or not. The first time series is the total hours from a YouTube channel over time and the second one is the organic hours (total hours - paid hours coming from ads). Both on a daily basis.

The total hours time series look like this:

enter image description here

After running the Augmented Dickey-Fuller test, the p-value is 0.038 (under 0.05), making the series stationary.

I was curious about how the mean and variance were behavouring over different sequences, plotting the following graphic. I can see most of the points are having a higher mean per each sequence compared with the global mean.

enter image description here

This is the plot for the second time series (total organic hours):

enter image description here

Then, I ran the ADF test to verify if it was stationary or not. The p-value is around 0.2191, which means the time series is not stationary.

Then, just to be sure, I plotted the variance and the mean for some sequences, obtaining the following:

enter image description here

I am curious why the first time series (total hours) is stationary when the mean seems to vary more than the second one (total organic hours), which is not stationary. Can this behavior be related to autocorrelation? Is it possible the first one is more autocorrelated than the second one, making a difference in the test?

Improve this question
$\endgroup$

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.