bug in Tables:Summary

[#1]

Sorry for the size of the file, but, 

Summarize by Column1, Statistics N for Data, Subgroup by Data and:

N for missing data is zero for all instances of Column 1, even though there are missing data.

This table was stacked from a mother file, which was a .csv, with missing data indicated by "nan".

Convert data to char and enter some value for missing data to see the real count.

[#2]

 I don't understand what you are doing with the subgroup by the data column, but the summary table indeed shows no missing data under N(Data,Missing) - but it does reveal missing data in the Data column.  So, I don't think there is a bug - I think you are mis-interpreting what column to look at to see the missing data.  For me, a more direct way to see missing data is to use the Missing Data Pattern platform rather than trying to explore it through a Table Summary.

[#3]

Like @dale_lehman , I am not sure why you are subgrouping by the data column. But if you are interested in the missing values, you could specify the N_Missing statistic. Even with data as the subgroup, it will show the correct number of missing values.

Ultimately, I think that subgrouping by the Data column confuses the issue and changes what you believe you are looking at.

[#4]

Forget the grouping variable, just look at the subgroup, and don't question why one would want to subgroup.  The sum of the subgroups, "Sum N(Data)", should add up to the number of items,"N Rows".   It doesn't because the 71,828 missing values are not being counted.  They should be.

 I'm not asking for help, I'm trying to report a bug.   What would be an appropriate venue for doing so?

N Rows  N(Data, Missing) N(Data, 0) N(Data, 1) N(Data, 2) Sum N(Data)
2007124 0                741019     801209     393068     1935296

[#5 · Accepted solution]

Hi, 

It seems that you have everything under control (no help needed). I would recommend contacting JMP support to report your experience (https://www.jmp.com/en_us/support/submit-a-problem-request.html?ctry=us_JMP)

Of note: this forum is used o exchange ideas and suggestions among average users

Best,

TS

[#6]

I think you are misinterpreting the report, as JMP is correct, I believe.

You want summary statistics for Data by Column 1, but with Data as a subgroup.

So, when you look at the report, the column labeled as 

N(Data,Missing) is telling you the number of data points in the data column, when the value of Data is missing. Well, there are MANY missing values in the data column. But when the data column is missing, guess how many rows of data exist? That's right, ZERO. that is the number you are seeing.

Similarly, your columns labeled as N(Data, 0) will tell you how many rows have data when the Data column =0. 

In my earlier comment I said that you would see the correct number of missing values if you specify the statistic N Missing. That column labeled as N Missing(Data, Missing) will give you the correct numbers because that will be the number of rows where the data is missing when the Data column is missing.

Sorry if this is confusing, but having a column named data can make that happen sometimes! I hope this helps.