How to Exclude Random Missing point before performing PCA.

[#1]

Hi,

How to exclude missing values from the dataset before analysis. These are random missing values due to instrument malfunction. I don't want to impute my data. I have explored the 'Explore Missing Value' but that didn't help because that changes the interpretation of my PCA results.

Is there any other way to do so?

Thank you,

Beginner_9

[#2]

How does your data look like? Can you exclude the whole row? If you can, select one of the missing values, right click on it, select matching cells and then right click on the row number and hide and exclude

You can also use Explore Missing Values for this. Select the columns with missing values you wish to exclude and then click Exclude Rows

[#3]

Thank you for your response. I knew the exclude option. I was looking for something where I can specifically exclude the random missing cells without having to apply imputation before my analysis.

Thanks alot

[#4]

Hi @Beginner_9,

You can use the Explore Missing Values platform to help detect, select and exclude rows with missing values automatically. It avoids having to search through your datatable and use manual operations.

Hope this may help you,

[#5]

Hi @Victor_G,

Is there a way to exclude missing cells instead of the entire row or column?

Thank you,

[#6]

Hi @Beginner_9,

Not with this platform, it will only give you the possibility to exclude rows containing missing values.
I suggest you read the documentation about the different methods available in the PCA platform.

If you want to "exclude" cells with missing values from the analysis, switching the default method to row-wise will exclude any observation with missing data. The other methods will try to leverage the information contained in missing and non-missing cells.

Hope this answer will help you,

[#7]

Hi @Beginner_9 ,

To exclude missing cells you can use Missing Value Codes. This excludes the value from one cell, but the values for the other variables are still included. You can use this also for outliers or noise as well and not just for missing values. You can see how at the Exploratory Data Analysis (around 2:40 min) Course from STIPS. This comes from the 2.6 Data Preparation for Analysis chapter. 

I hope that solves the problem,
Spyros