How to seperate date format 20170203T222637+0000

[chenzinc]

i am validating a huge dataset where it has 8000 observations, may I know how can I extract the date from this format (20170203T222637+0000)?

what i want: 20170203

what i have: 20170203T222637+0000

is there another way to extract it at once for 8000 observations?

[Kurt_Bremser]

Is this a character variable, or numeric with a B8601DX format assigned?