Hello, I am trying to do hierarchical cluster analysis, from a .csv file, with Rstudio but I have NULL values. Is there a library or even a parameter that I can use so I can do cluster analysis without the NULLs?
Well, if the NULL means that it's a missing value, then you just need to use a missing value strategy, where you either 1) throw away the tuple, 2) replace it with the most common value, or 3) replace it with the average of the other values. There are some other strategies, but that should at least get you started.
You can go through techniques of various missing values handling approaches
You have to determine whether the missing values are random or not and if they are significant enough so that the error of trying to model a given distribution would be off significantly.
Some values can be handled by imputation because of underlying missing predictors. For example, people who have financial distress can be less likely to disclose their financial status than someone who is well economically and likes to brag. Given that low and middle income families are less than rich people, if your data does not reflect this then you can(being very careful) model and impute the missing values.
The ability to determine if imputation is appropriate or not is a decision that you have to take into account carefully through analysis of the data.
- Badges
- Science topic
- Similar topics
- Correction