Misc. Stats Topics Discussion > PCA using Statitica
Dear Siladitya,
Unfortunately, I'm not familiar with Statistica as a software platform. However, that type of error message in most programs would usually indicate that you have too much missing data on the items that you are using for the analysis. A "valid case" in most statistical software platforms is typically a case with a specified amount of non-missing data (if the analysis is using listwise deletion, then a valid case is one with no missing data; if the analysis allows for some missing data, such as pairwise deletion or MLE, then a valid case is likely one that isn't missing all values for the variables). It is most likely that Statistica is considering valid cases as not missing any data for the key variables in this case, but I can't say for sure without knowledge of the program.
Here are a few resources from the web on Statistica, though they seem very basic in nature:
StatSoft Video Tutorials for Statistica:
http://www.statsoft.com/support/download/video-tutorials/
U of Toronto Brief Tutorial:
http://www.psych.utoronto.ca/courses/c1/statistica/page1.htm
Thank you for your reply. But in a 16x44 matrix, I have only 6 cells where the value is 0. Is it the reason? Very frankly, I do not know. Will it improve, if I remove all those variables with even single 0 value? Please suggest.
Siladitya,
I wasn't referring to values of zero, but missing values. With that said, I can't be sure that is the problem, since I'm not familiar with Statistica. Sorry. I hope the resources I provided are helpful.
While trying to do PCA using Statistica, I am receiving the following error message:
The percentage of valid cases (65.67%) for variable files is less than the required minimum percentage of valid cases (80.00%). The analysis stops.
How to solve this issue?
I have 40 variables which I want to compare on the basis of their presence in 44 files, the names of which I kept at the last column as a separate variable.
Please help.