Why topic modeling discovers themes in NLP - Test Your Understanding

Practice - 5 Tasks

1fill in blank

easy

Complete the code to import the library used for topic modeling.

NLP

from sklearn.decomposition import [1]

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

APCA

BLinearRegression

CKMeans

DNMF

Attempts:

3 left

💡 Hint

Common Mistakes

Choosing PCA which is for dimensionality reduction but not typically for topic modeling.

Choosing KMeans which is a clustering algorithm, not matrix factorization.

The NMF class from sklearn.decomposition is commonly used for topic modeling to discover themes in text data.

2fill in blank

medium

Complete the code to fit the topic model on the document-term matrix.

NLP

model = NMF(n_components=5, random_state=42)
W = model.[1](X)

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

Atransform

Bfit_transform

Cfit_predict

Dpredict

Attempts:

3 left

💡 Hint

Common Mistakes

Using transform without fitting first causes an error.

fit_predict is not a method of NMF.

The fit_transform method fits the model and returns the transformed data (topic weights for documents).

3fill in blank

hard

Fix the error in printing the top words for each topic.

NLP

feature_names = vectorizer.get_feature_names_out()
for topic_idx, topic in enumerate(model.components_):
    print(f"Topic {topic_idx}:", end=" ")
    print(" ".join([feature_names[i] for i in topic.argsort()[1]()[:10]]))

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

A::-1

Bascending

Creverse

Dsort

Attempts:

3 left

💡 Hint

Common Mistakes

Using sort() which returns None and causes an error.

Using reverse() which works only on lists, not numpy arrays.

Using argsort()[::-1] sorts indices in descending order to get top words per topic.

4fill in blank

hard

Fill both blanks to create a dictionary of topics with their top words.

NLP

top_words = {f"Topic_{i}": [feature_names[j] for j in model.components_[i].[1]()[2][:5]] for i in range(model.n_components)}

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

Aargsort()[::-1]

Bsort

Creverse

Dargsort

Attempts:

3 left

💡 Hint

Common Mistakes

Using sort which returns None.

Trying to use reverse which is not valid here.

We use argsort()[::-1] to get indices of top words in descending order. The second blank is empty because slicing is done after reversing.

5fill in blank

hard

Fill all three blanks to compute and print the coherence score of the topic model.

NLP

from gensim.models import CoherenceModel
texts = [[word for word in doc.split()] for doc in documents]
coherence_model = CoherenceModel(model=model, texts=texts, dictionary=[1], coherence=[2])
coherence_score = coherence_model.[3]()
print(f"Coherence Score: {coherence_score:.3f}")

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

Adictionary

Bcorpus

Cget_coherence

Dc_v

Attempts:

3 left

💡 Hint

Common Mistakes

Passing corpus instead of dictionary.

Using wrong method name to get coherence.

The dictionary is needed for coherence calculation, c_v is a common coherence type, and get_coherence() computes the score.