A question about standard deviation for drawing

[shijiwensjw]

https://github.com/alexlee-gk/video_prediction/blob/master/scripts/plot_results.py#L25
In this line, I wonder why you use metric_se = np.std(metric, axis=0) / np.sqrt(len(metric)) instead of just standard deviation metric_se = np.std(metric, axis=0) ? Why is the standard deviation divided by np.sqrt(len(metric)?

[richzhang]

It's the standard error -- the more measurements you make, the more sure you are of the mean.

[shijiwensjw]

Thank you, richzhang,
I confused the standard deviation and the standard error, now I understand.