Workshop: Intro to topic modeling using Python

on YouTube ↗01h 01m

" summary

Uses Python with spaCy preprocessing, scikit-learn vectorizers and K‑Means, and gensim LDA on a 3% Yelp sample (~31,500 reviews). K‑Means on 1,000-term TF‑IDF vectors runs in minutes, yields five initial clusters but a silhouette peak at 20 clusters, with one cluster dominating >50% of documents. Gensim LDA returns topic–word probabilities, finds LDA coherence favors 5 topics, reports 37 distinct top words (vs 28 for K‑Means) and 21 words in common; PCA plots show partial separability.

AI-written summary · about

" chapters (16)

1. 0:00Start
2. 5:07What Is Topic Model
3. 7:43Topic Modeling
4. 8:13Generative Model of Topic Modeling
5. 11:34Tokenization
6. 17:09Sparse Matrix
7. 17:34Measure Similarity
8. 18:14Apply K-Means Clustering
9. 21:55Word Counts
10. 29:19K-Means Clustering Method
11. 30:32Mean Nearest Cluster Distance
12. 30:54Silhouette Coefficient
13. 38:43Training Process
14. 51:34Identify the Actual Language of a Review
15. 54:14Word Frequencies
16. 55:04Common Top Words

" related talks

• 48mTopic Modelling (and more) with NLP framework GensimBhargav Srinivasa Desikan · PyData Berlin 2017
• 01h 31mModern NLP in PythonPatrick Harrison · PyData DC 2016
• 45mDecoding Topics: A Comparative Analysis of Python’s Leading Topic Modeling Libraries Using Climate CLisa Andreevna Chalaguine · PyCon DE & PyData 2025
• 45mEvaluating Topic ModelsMatti Lyra · PyData Berlin 2017
• 02h 19mWorkshop: Demystifying Natural Language Processing using Python (Scikit-Learn/ Keras)Vaibhav Srivastav · PyCon APAC 2019
• 49mTopic Modeling: A Provable Spectral MethodRavi Kannan · Spectral Algorithms: From Theory to Practice 2014
• 01h 19mGenerating Customers Insights with Topic Modelling and HuggingFace SetFit MethodSultan Al Awar · PyData London 2024
• 35mAnalyzing Public Conversation using LDA and Topic Modeling, in PythonSveta Gimpelson · stackconf 2023