Transformations

andyw

New Member
#1
I have two datasets, before and after the application of a treatment. The treatment is theorised to produce an increase in "factor x", and difference between the two datasets should be a positive value, however due to noise/error etc in the data some of the difference values are negative.

In order to assess the "difference data" I need to transform it in some way. Was planning on doing a log transformation, however this causes problems with the negative values.

Any ideas on how best I could adjust the data for log transformation, or an alternative transformation that could be used?

Thanks,

Andy
 

noetsi

No cake for spunky
#4
It seems that you are trying to apply a filter to eliminate some extraneous factor to get at the true relationship between the tables. If that is the case you need to find a transformation that will actually do that, likely in the literature on this topic for from someone who works in the area. Without knowing specifically (not generally like "eliminate noise") you won't be able to find the correct transformation.