How can I convert 2D matrix into a sequence?

I need to convert a matrix into a sequence. I have performed PCA on it and obtained 6 principle components. The PC1 contains almost 51 percent information. PC1 and PC2 contains 99% information. Is there any way that I can use PC1 and PC2 as a sequence in a sequence classifier or generate a sequence from above mentioned matrix. Anyone can give any hint or idea please.

[Jonas]

for PCA reconstruction, look into here. Also have a look into the answer's comments

[Rab Nawaz]

@Jonas, Thanks for your comments, I check it out.

Answer 1

[Purvaja]

Hi @Rab Nawaz,

Assuming the data to be time series or sequential data, we can depict the matrix data as sequence data as follows:

Using the 1800x2 PCA Matrix (from Principal components PC1 and PC2)

1.Treat each row (i.e. one time step of PC1 and PC2) as an input to your sequence model:

% Example PCA data 
PCA_data = rand(1800, 2);  % Dummy data 

% Convert to sequence: each row becomes a 1×2 sequence 
sequenceData = mat2cell(PCA_data, ones(1800, 1), 2); 

Output: a 1800×1 cell array, each cell is a 1×2 vector like [PC1, PC2].

You now have a proper input format for sequence models, where each time step is a feature vector [PC1, PC2]

2. Add labels as necessary (for classification).

labels = categorical(randi([1, 2], 1800, 1));  % Dummy binary labels 

3.Train on any sequence classifier like LSTM.

For more clarifications refer to the following official MathWorks documentation links:

1. PCA: https://www.mathworks.com/help/stats/pca.html
2. Sequence classification using deep learning: https://www.mathworks.com/help/deeplearning/ug/classify-sequence-data-using-lstm-networks.html

Hope this helps you!