Categories
Algorithms OMSCS

How To Compute Reconstruction Error for Random Projection

Random Projection is an interesting Dimensionality Reduction technique. You may choose to create a random projection for 1,2,3,..,n dimensional projections. Now how to tell which one is best?

So you would need to calculate loss of data due to this reduction in data size.

# data has this shape:  row, col = 4898, 11 
random_projection = SparseRandomProjection(n_components=5)

random_projection.fit(data)
components =  random_projection.components_.toarray() # shape=(5, 11) 
p_inverse = np.linalg.pinv(components.T) # shape=(5, 11) 

#now get the transformed data using the projection components
reduced_data = random_projection.transform(data) #shape=(4898, 5) 
reconstructed= reduced_data.dot(p_inverse)  #shape=(4898, 11) 
assert  reduced_data.shape ==  reconstructed.shape
error = mean_squared_error(data, reconstructed)

Leave a Reply

Your email address will not be published. Required fields are marked *