2
$\begingroup$

My question is whether the use of DimensionReduce, for a simple example with Method -> "PrincipalComponentsAnalysis" is the equivalent to non-detect PCA, i.e. PCA for data, some of which are known only as an interval.

For example

data = RandomVariate[MultinormalDistribution[{3, 3, 3}, 
{{1, -0.5, 0.5}, {-0.5, 1, -0.25}, {0.5, -0.25, 1}} ], 1000];

eData = EventData /@ (data  /. {{_?(# < 1 &), y_, z_} -> {{0, 1.}, y, z}, 
{x_, _?(# < 1 &) , z_} -> {x, {0, 1.}, z}});

redeData = DimensionReduce[eData, Method ->  "PrincipalComponentsAnalysis"];
redData = DimensionReduce[data, Method ->  "PrincipalComponentsAnalysis"];
ListPlot[{redData, redeData}]

Gives the result as

ReduceDimense for original data & EvenData

Regardless undetected data is only a small part of the file.

$\endgroup$
  • $\begingroup$ Can you please edit your question to clarify what is that you are asking? $\endgroup$ – rhermans Jun 28 '18 at 19:27
1
$\begingroup$

(Extended comment.)

It appears that DimensionReduce on the event data eData is going to use a feature extractor that puts eData in an 8-dimensional space, and then the reduction is done over that 8-dimensional representation.

enter image description here

Compare with the same results for data. It appears that data is not changed before the reduction.

enter image description here

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.