How to use data after the dimensionality reduce for classification

Hello.

I have a dataset that applied dimensionality reduce like PCA.

I attached the file. The dataset is consisted of 120 x 2353 (column 2353 is label, 0~6).

How can I use these dataset for classification?

Answer 1

[Image Analyst]

You can take a certain number of PCs and threshold them. For example, you have class 1 if PC1 < 0.5 and PC2 > 0.8 or whatever. It would help if you could visualize your PC's via a scatterplot or image or something so you can see what really matters. Or you could get Eigenvector's PLS Toolbox which has extensive and very sophisticated tools for figuring out your question.

[Kong]

Hello.

I am wondering how to use the dataset applied dimensionality reduction.

For example,

the data 1 (70 x 90) is related to label 1,

the data 2 (70 x 90) is related to label 2,

the data 3 (70 x 90) is related to label 3,

To use the classification of these data, I will flatten each data.

the data 1 (70 x 90) is flattened to 6300, the data 2 (70 x 90) is flattened to 6300

Finally, I get the full dataset like 3 x 6300.

Is it reasonable to convert the dataset like this? I am curious that this process may lose the property of the dataset.

[Image Analyst]

Yes, it's what you should do. This is similar to doing PCA on an RGB image where you have three 2-D color channels. See attached demos.