Statistics in Diadem on a block of data

[RussellSenior]

I am trying to calculate some statistics on a block of data. Specifically, the standard deviation.  I have 8 channels, each with 300 rows of data.  When I use the descriptive statistics, I am not getting the standard deviation for the entire population, I am only getting it for the first channel, or the first row (if row oriented evaluation is set). 

So, if I put all 8 channels into the channels box of the Descriptive Statistics GUI, I only get the standard deviation of the first channel.  If I use Row-oriented evaluation, I only get the standard deviation for the first row of data. I verified this by copying the data to excel and using the sample standard deviation calculation.  

How can I force Diadem to give me the standard deviation for the entire population? 

[RussellSenior]

Here is a data set to play with.  

If you pull the data into excel, the expected sample standard deviation for all the data is 0.222353 bar.  However, Diadem reports 0.161bar, which is the standard deviation of only "Pressure Data 1" as reported by excel. It works this way for the Arithmetic mean as well. 

I have Rows set to "1-" 

The only workaround I can think of is to concatenate my data into one channel, but that is quite inconvenient. 

[RussellSenior]

Ok, so concatenating all the data wasn't as painful as I thought it would be... 

What is the use of being able to specify a block of data to the statblockcalc function if the block of data isn't actually used for analysis?  Or am I doing something wrong? 

[Brad_Turpin]

Hi Russell,

The block of data that the statistics are calculated over is actually a row subset from each channel specified.  The statistics calculation is one of a number of ANALYSIS dialogs in DIAdem that enable you to run calculations on multiple channels with one press of the <Calculate> button.  The behavior of such ANALYSIS dialogs is to implicitly loop the calculation over each channel specified.  The calculation is being carried out on only one channel at a time, in isolation from all the other channels.  The default is to consider all the data in a given channel when calculating its statistics.  If you specify a row subset other than "1-", though, then you will be calculating the statistics for that channel using only the data in that row range.

Brad Turpin

DIAdem Product Support Engineer

National Instruments

[RussellSenior]

So, if I have 4 channels (columns) of data, each with 100 elements (rows), how do I do statistical analysis on all 400 elements?  

[gsklyr]

Each channel is a set of data.  You can do statistical analysis on each channel separately and get the results as properties of that channel.  If you want to preform the same calculations on the data of all 4 channels as 1 data set, you will have to concatenate all 4 channels into 1 channel, and then run your statistics on that channel.  The equivalent in Excel would be to put all 400 values in one column.

To concatenate channels use:

Call ChnConcat("Group/SourceChannel","Group/TargetChannel")

"