How to know number of features and set up the size of image matrix?

[utkarsh0902311047]

Hi, Can you help me how to know how many features are going to be used after imagetransformer and which are those features? Also how can i set up the size of the image, lets say I have 121 features and I need 11x11 image matrix. Actually I am looking for a method which can convert every row in my data frame into a nxn matrix image.

Thank you Utkarsh

[kaboroevich]

All features will be still be present and assigned to a pixel after image transformation. After training the image transformer, you can extract the pixel coordinates of the features using .coords() as a two column dataframe. The index of that dataframe will match the order of the input data.

However, I do not think DeepInsight is the correct method for you if you want to uniquely assign each feature to a pixel. DeepInsight will result in some overlapping features (features assigned to the same pixel) as well as empty pixels. I think reshaping your matrix rows is what you want. For example, if you have a numpy array of 10 rows and 121 features:

import numpy as np
a = np.random.rand(10, 121)
b = a.reshape(10,11,11)

b will be a numpy array of 10 11x11 matrices.