How to calculate multiple means if dataset is irregular

JoenStat

New Member
For a longitudinal study I'm building a database, I cut down 24 hours measurements in 96 mean value's of 15 minutes. The raw data is is irregular, i.e. it is measured every 5 or 6 seconds. (see attached picture) Anyone has an idea how to automatically calculate 15 minute mean values of this dataset?

I prefer to use excel or SPSS. Thanks in regards.

Attachments

• 39.9 KB Views: 2

hlsmith

Less is more. Stay pure. Stay poor.
Are you expecting to see a change in means across time? I would imagine if you have an increment that spans two 15 minute periods and has 3 seconds in one group and 2 seconds in the other that it would not be a huge deal in which one it ends up in correct, since it is just 1 of 60 obs used in calculation and it will be accounted for in the mean denominator.

Given this I would just break the data up into 15 min increments and calculate means accordingly. Some may have 59 or 61 obs per period, but this will be negligible.

JoenStat

New Member
Are you expecting to see a change in means across time? I would imagine if you have an increment that spans two 15 minute periods and has 3 seconds in one group and 2 seconds in the other that it would not be a huge deal in which one it ends up in correct, since it is just 1 of 60 obs used in calculation and it will be accounted for in the mean denominator.

Given this I would just break the data up into 15 min increments and calculate means accordingly. Some may have 59 or 61 obs per period, but this will be negligible.
Problem is I have many measurements of 24 hours, so it is too much work to break them all into 15 minutes piece manually. I’m searching for a way to cut the data into 15 minutes pieces automatically, then it’s no no problem they are based on 59 or 61 observations per 15 minutes.

I thought of imputing the data so that there’s one observation every second. Then the data will be regular. Problem is I don’t know how to impute data when there are no empty rows.

hlsmith

Less is more. Stay pure. Stay poor.
Got a little lost, can't you just figure out the best denominator and instead of finding say quartile you find ?? percentiles to split data into groups?

JoenStat

New Member
Got a little lost, can't you just figure out the best denominator and instead of finding say quartile you find ?? percentiles to split data into groups?
The problem is that I have many different datasets with different missing observations. I still have to manually split the data