I have a dataset with 2 numerical fields (X and Y) and one categorical field (only two types).
If I use a scatterplot I can plot my data with two different symbol for each type (see the figure).
I'm looking for the best equation able to split/separate my data in two group.
My deal was to use R so R code to be able to do that could be helpful.
Thanks
You are looking for unconstrained classification with mixed types. A lot of techniques could be used. The scikit-learn (a machine learning Python library) team has shipped an interesting flow chart to help selecting the appropriate technique. Most of these techniques (if not all of them) are also available in R (look for the 'caret' package). You might look for techniques able to handle mixed types (e.g. clustering based on a Gower distance matrix) to avoid transforming your categorical field through an risky numerical transformation. You might also scale your data (normalization, standardization, scale to a range, etc.) before applying the technique.
http://scikit-learn.org/stable/tutorial/machine_learning_map/index.html
http://caret.r-forge.r-project.org/
Hi all,
I understood that the categorial variable indicates group membership, and that an equation or a model was sought to describe how the data can be assigned to one of the groups based on the two variables X and Y. My first idea was discriminant analysis, either using all or only part of the available data to estimate the parameters of the discriminant function. However, your diagram reveals that there is a lot of overlap between the two groups, so I do not expect the separation to be really effective...
Cheers
Tobias
Not sure if I got your problem, but wouldn't the function split() do the job?
Let df be your data.frame with columns X, Y, A (A is the factor). Then
split(df, by=df$A)
returns a list with as many items as there are levels in df$A (=two in your case). So all the parts of df got separated according to the levels in df$A.
If I understand the question, you are trying to physically just separate the data? The above answer would work but I can also recommend the following:
Say your data is called data and your factor is called factor with levels A and B:
subset1 = subset(data, factor=="A")
you could also specify variables to extract from the data frame by specifying a select parameter in subset:
subset1 = subset(data, factor=="A", select=c(var1,var2,var3))
I suggest you read the article: http://www.researchgate.net/publication/278403191_Numerical_Coding_of_Nominal_Data
Article Numerical Coding of Nominal Data
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R Statistical Package