How to predict map (gridded raster) after training and fit a machine learning model?

Hi, The task is to do the classification from satellite imagery. I already did the following steps:

0. Prepare a Landsat 8 scene (4 bands), a ROI polygon.

1. Create the ROI and export to CSV file.

2. From ROI (csv file), XTrain, YTrain, XTest, YTest were made by 70% for training.

3. Trained and fitted a machine learning model (random forest, CCF).

4. Predicted from the test data (XTest) and saved in the variable Ypred.

Now, I would like to do the prediction for the whole image (with X will be Landsat 8 4 bands) to have the classified map.

My question is: How will I do this task in Matlab?

This can be done in R and Python. But my friend give me a piece of code in Matlab format and I start again from sratch. I found an article here:

https://au.mathworks.com/matlabcentral/answers/386232-how-to-convert-an-array-to-an-image

but seems not work for my case.

Any comments and links to useful docs will be very appreciation.

Thanks, Thang