Latent Dirichlet Allocation (LDA) in NLP - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to import the LDA model from scikit-learn.

NLP

from sklearn.decomposition import [1]

Drag options to blanks, or click blank then click option'

AKMeans

BPCA

CLatentDirichletAllocation

DTruncatedSVD

Attempts:

3 left

💡 Hint

Common Mistakes

Importing PCA or KMeans instead of LatentDirichletAllocation.

Misspelling the class name.

The correct import for Latent Dirichlet Allocation in scikit-learn is LatentDirichletAllocation.

2fill in blank

medium

Complete the code to create an LDA model with 5 topics.

NLP

lda = LatentDirichletAllocation(n_components=[1], random_state=42)

Drag options to blanks, or click blank then click option'

A5

B3

C10

D1

Attempts:

3 left

💡 Hint

Common Mistakes

Choosing a different number of topics than requested.

Confusing n_components with other parameters.

Setting n_components=5 creates an LDA model with 5 topics.

3fill in blank

hard

Fix the error in the code to fit the LDA model on the document-term matrix named 'dtm'.

NLP

lda.fit([1])

Drag options to blanks, or click blank then click option'

Adocuments

Bdtm

Ctopics

Dlabels

Attempts:

3 left

💡 Hint

Common Mistakes

Passing raw documents or labels instead of the document-term matrix.

Using a variable that does not exist.

The LDA model fits on the document-term matrix, which is named dtm here.

4fill in blank

hard

Fill both blanks to get the topic distribution for the first document.

NLP

topic_distribution = lda.[1]([2])[0]

Drag options to blanks, or click blank then click option'

Atransform

Bdtm

Cfit_transform

Ddocuments

Attempts:

3 left

💡 Hint

Common Mistakes

Using fit_transform after fitting the model.

Passing raw documents instead of the document-term matrix.

Use transform on the document-term matrix dtm to get topic distributions. Index 0 gets the first document's distribution.

5fill in blank

hard

Fill all three blanks to create a dictionary of top words per topic from the LDA components.

NLP

top_words = {i: [feature_names[j] for j in lda.components_[i].argsort()[-[1]:][::-1]] for i in range([2])}
print(top_words)

# feature_names = vectorizer.get_feature_names_out()
# lda = LatentDirichletAllocation(n_components=[3])

Drag options to blanks, or click blank then click option'

A10

B5

Clda.n_components

Dlda.components_.shape[0]

Attempts:

3 left

💡 Hint

Common Mistakes

Using a wrong number of top words like 10 instead of 5.

Confusing the number of topics with the number of words.

Using a fixed number instead of dynamic attributes.

We get the top 5 words per topic, so 10 is too many. The number of topics is the number of rows in lda.components_, which is lda.components_.shape[0]. The n_components parameter is the number of topics set in the model.