Coding truncatedSVD with scikit-learn

Using scikit-learn can help us to understand m, A, U, Σ, V T. The main class is truncatedSVD(). Let’s assume matrix A is a 5 x 10 matrix. In LSI, it means there are 5 documents and 10 unique words. Let’s fill in random integer values between 1 and 50 (low = 1 and high = 50):

import numpy as npA = np.random.randint(low=1, high=50, size = (5,10))
print(A)

The output looks like this:

[[ 2, 23, 38, 24, 32, 20, 22, 38, 4, 6],[35, 20, 47, 49, 29, 39, 15, 15, 8, 28],
[35, 8, 47, 2, 40, 24, 21, 37, 12, 25],
[43, 41, 22, 41, 27, 45, 41, 31, 36, 28],
[19, 17, 8, 39, 40, 24, 43, 16, 33, 22]]

We will decompose matrix A using TruncatedSVD.

Using TruncatedSVD

The TruncatedSVD function of sklearn.decomposition takes an input parameter, n_components. Let’s declare the TruncatedSVD object (known as svd) and assume there are three topics (n_components=3). I will explain n_components later:

from sklearn.decomposition...