1. ## Mean comparison

Hi everyone.
I am sorry if this is too simple question, but I still cant deal with it.
I have a set of data for period of time. I know that this data is going to change in different months and my task is to find by how much. I am calculating mean values for each month and compare those means to mean of whole set of data (for all of the period). In the result I got set of deviations for each month. I am a little bit confused because sum of deviations do not give zero. As I understand, this is caused by different amount of data points in different months.
Are my calculations correct so far or I need to use some king of weights?

2. ## Re: Mean comparison

hi,
there are many ways you could solve this problem, the choice should depend on the goal of the measurement. Do you want to get an alarm if the change is above a limit or do you just want to visualise the evolution of the system or something else entirely?

regards

3. ## The Following User Says Thank You to rogojel For This Useful Post:

Alergenic (12-13-2016)

4. ## Re: Mean comparison

Originally Posted by rogojel
hi,
there are many ways you could solve this problem, the choice should depend on the goal of the measurement. Do you want to get an alarm if the change is above a limit or do you just want to visualise the evolution of the system or something else entirely?

regards
the goal is to find how does average value change each month. We can take an average value of each month average and compare them, then the sum of deltas will be zero. Is it correct?

And if we have some number of data sources, and we make the same analysis for each source. Can we take an average value of those deltas (average for January from different sources, here we assume that there is correlation in different source data month to month)? Should we make some weightening here? I mean there is different number of data points for each month inside one data source and there is different number of data points for each data source.

5. ## Re: Mean comparison

I think that deviations of the monthly averages should not sum to zero, just as you said, due to the varying number of points that come into each average. Now, if you just want to find month that deviate and maybe check for an effect of different data sources then a two-way ANOVA might be the way to go. Or, alternatively, a control chart per each data source - e.g. an Xbar-R could give you a good description of whether it is business as usual or something might have happened. It really depends on your goal..

 Tweet

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts