Function for visualisation of raw data -Ovarian cancer

I want to a 'plot' function to effectively visualise the ovarian cancer data inbuilt in MATLAB.

load ("OvanrianCancer.mat")

Error using load
Unable to find file or directory 'OvanrianCancer.mat'.

The variable X (a 4000 by 216 matrix containing 4000 gene expression from 216 patients with and without ovarian cancer) is what I want to visualisation in its raw form to get an insight of its patterns. I intend to use a modal decomposition algorithm to analyse and re-construct it.

I have tried different 'plot' function but the results are awlful. I guess it's me not using the right 'syntax' or commands.

[Walter Roberson]

What do the entries look like? For example, what are the first 10 entries of the first row?

[Cutie]

Each column contains 4000 gene expression samples of a patient.

[Image Analyst]

Please attach 'ovariancancer.mat' with the paperclip icon.

[Walter Roberson]

surf(obs)

??

[Cutie]

@Walter Roberson surf(obs) didn't give anything meaniful

@Image Analyst I have attached the data

Answer 1

[Image Analyst]

Try this:

load('OvarianCancer.mat')
plot(X, '-');
grid on;

Do you want to make a model of it? Do you have any ground truth for the data?

[Cutie]

i did this plot function before but the confusion is that the visuals are too small to make insights from. I use semilogy.

I am analysing the data using a modal decomposition algorithm. My model is to be able create a reconstructed replica of the original data (with its patterns/structures).

Maybe, some 'manipulation' ideas to perfect the plot functions can help. e.g setting the limits.

[Walter Roberson]

You have over 800000 data points of floating point values that you are asking to draw in a single graph that probably is only about 1000 pixels + borders wide. It is not obvious to me that you can have any possible meaningful visualization — not unless you are doing some kind of summary like histogram.

Answer 2

[Jeremy Huard]

This is a preprocessed Mass Spectrometry dataset. plot(X) will indeed show the spectra but you'll need data processing including dimensionality reduction to gain meaningful insights.

For example, check this example that uses this exact dataset: Select Features for Classifying High-Dimensional Data