Data Science: It's Easy as Pyǃ
summary
Demo builds a language classifier from Wikipedia articles: pages are fetched via curl/uncurl, cleaned with wikipedia-extractor plus a regex, and analyzed in pandas and seaborn. Character tf-idf character n-grams (bi/trigrams) vectorize text, an SVC with pipeline trains on label-encoded language codes, and PCA plus k‑means visualize clusters. RandomizedSearchCV tunes hyperparameters. Initial accuracy is ~96%, merging near-identical South Slavic labels raises accuracy above 99%, showing the approach is fast, reproducible, and practical for language detection.
AI-written summary · about
chapters (26)
- 0:00Intro
- 1:40What I'll Cover
- 2:09What is Data Science?
- 3:31OSEMN!
- 5:03Language Classification
- 6:43Scrape All The Things!
- 8:11Scrub
- 8:57Pandas
- 15:09Machine Learning
- 16:29Visualize
- 19:39KMeans Clustering
- 22:19Explore More
- 23:23Explore Even More
- 24:38Supervised Learning
- 25:42The Model
- 31:57X_train
- 32:49Tuning
- 33:11Randomized SearchCV
- 34:26Skull
- 34:41sklearn.metrics
- 35:13Overfitting
- 35:52accuracy_score
- 36:27classification_report
- 37:30I'm Confused
- 39:29y_train
- 39:48INterpret
related talks
01h 39mNatural Language Processing (NLP) in Python - From Zero to Hero
01h 17mOne workshop that data scientists don't want you to attend...
30mKick-Start your Understanding of Machine Learning with Python
01h 17mMachine Learning with Scikit-Learn
48mIntroduction to Natural Language Processing with Python
34mApplying Machine Learning Techniques to Detect and Understand Fake News
43mEveryone can do Data Science in Python
34mSolving Crimes with Python Using Natural Language Processing (NLP)