Feature Engineering is for Everyone!
summary
Claims that non‑scientists can create features get tested with runnable Colab notebooks using pandas, ydata‑profiling, scikit‑learn, CountVectorizer, Yellowbrick, sentence‑transformers and yfinance. The session shows one‑hot encoding and label encoding, builds TF‑IDF vectors, runs cosine similarity and a nearest‑neighbors search to find surprisingly similar review pairs, and produces plots and rolling‑window, min‑max and log transforms for stocks. Notebooks execute end‑to‑end but require sampling on Colab because of RAM limits.
AI-written summary · about
chapters (44)
- 0:00Introduction to feature eng
- 1:11Workshop overview and use case
- 3:48Google Colab setup and tools
- 6:29Exploring discrete data
- 8:45Data loading and profiling
- 11:42Discrete data profiling exercise
- 15:27Visualizing review titles
- 17:20Text feature creation exercise
- 18:39Profiling report review
- 23:26Encoding discrete data
- 26:08Lexical vs semantic similarity
- 29:47TF-IDF and embeddings
- 32:47One-hot encoding in Python
- 36:06Label encoding in Python
- 39:00Word counts and n-grams
- 44:03TF-IDF implementation
- 52:11Text similarity exercise
- 57:33Sentence transformers
- 1:04:24Nearest neighbors analysis
- 1:07:43Similarity exercise analysis
- 1:14:34Data engineering exercises
- 1:18:01Nearest neighbors implementation
- 1:21:45Visualizing similarity results
- 1:24:10Continuous data exploration
- 1:25:58Stock data visualization
- 1:28:13Data profiling for stocks
- 1:32:27Box plots for distribution
- 1:35:00Heat map visualization
- 1:37:17Scatter plots for correlation
- 1:39:13Stock data exercise
- 1:40:53Profiling report walkthrough
- 1:43:49Continuous data transformation
- 1:45:47Log transformation example
- 1:48:36Data ratios and relationships
- 1:50:07Rolling windows and trend smoothing
- 1:52:19Lags and time relationships
- 1:53:26Date time feature extraction
- 1:55:25Data Science Rebalanced workshop
- 1:57:13Normalization in Python
- 2:00:54Log transformation in Python
- 2:03:24Relative strength ratios
- 2:05:04Rolling average implementation
- 2:07:39Visualizing rolling trends
- 2:10:24Time series lag analysis
related talks
39mFeature Engineering: Unlocking Potential for Predictive Models
01h 28mFeature Engineering Made Simple
30mFeature Engineering: An Apprentice’s Guide to the “Dark Art” of Machine Learning
45mAnatomy of a production ML feature engineering platform
28mData VS Engineers: a long lasting story
51mSelf-serve Feature Engineering Platform Using Flyte and Feast
17mCreating & operating ML models from event-based data using feature stor...
01h 17mOne workshop that data scientists don't want you to attend...