principal component analysis in LabVIEW

[alexderjuengere]

so, you want to do "leak detection"

does this mean your final result is a dual classification like 

case1 ="there is a leak detected"

case2 = "there is no leak detected"

A PCA would help you "to show" the difference between case1 and case2, but the difference itself is within the data!

[alkamohan]

ya.that's what am really supposed to do.but am in a dilemma.i have the leak signal.i obtained its parameters(16 of them like mean,variance etc)under 10 differnt conditions..now upon doing PCA what would it really do?

[alexderjuengere]

1#

Ok, you have achieved quite a lot:

- you acquire data from sensors,

- you normalize the data,

- you extract certain features (mean, variance, etc.),

- you know how to do a PCA and you know what a PC is.

But did I get this right, you did just take the leak signals into consideration?

Have you already checked if there is a difference in the feature vector space between non-leak and leak signals?

If not, you maybe have to extract different features from your (normalized) data.

2#

Regarding your .xls files:

Does a row represent a feature like row 1 = mean, row 2 = variance, .... ?

or does each row represent one measurement?

If each row represents one measurement, which one belongs to class leak signal and which one belongs to class non-leak signal?

Alex

[alexderjuengere]

hi,

maybe you've already found  a solution for your task, but just in case not, here's a toolkit which only requires Labview Standard functions to do PCA and KernelPCA:

https://decibel.ni.com/content/docs/DOC-19328

2.3 Dimension Reduction Algorithms

Dimension reduction refers to the process of reducing the number of dimension of the data. The projection of the data set in the reduced space is often desired to preserve certain important data characteristics. In some cases, data analysis, such as clustering, can be done more easily and accurately in the reduced space than in the original space. One prime application of dimension reduction is face recognition, where face images represented by a large number of pixels are projected to a more manageable low-dimensional feature space before classification.

List of functions:

• [...]
• Principal Component Analysis (PCA)
• Kernel PCA
• [...]

Alex