Sculpting Data for Machine Learning V02
summary
Three curated datasets are released on Daggle: an e‑commerce size/feedback set (userID, productID, size, fit, measurements), a sarcasm corpus combined from The Onion and Huff Post, and a Huff Post news‑category corpus; collection uses guided vs unguided search and a Pythonic scraping/ETL pipeline with open‑source tools. Workflow emphasizes data trimming, cleaning inconsistent sizes/dates/spelling, vertical and horizontal integration, and focused feature engineering, claiming that better data and features can boost model performance by up to 100X.
AI-written summary · about
chapters (12)
- 0:00Start
- 0:33Jigyasa Grover
- 7:02Sarcasm Detection Data Set
- 8:03News Category Data Set
- 8:57Searching for the Data Sources To Build Good Quality Data Sets
- 9:39Guided Search Approach
- 10:23Defining the Problem Formally
- 12:28Data Volume Consideration
- 21:25Data Integration
- 24:27Feature Engineering
- 25:57A Typical Machine Learning Workflow
- 26:48Data Centric Ai
related talks
30mData-Centric AI Cookbook-let's prep that data
11mRethinking Dataset Discovery with DataScout
44mData Journalism 1
01h 17mOne workshop that data scientists don't want you to attend...
29mSeveral People Are Tuning: Data and Machine Learning at
01h 19mGenerating Customers Insights with Topic Modelling and HuggingFace SetFit Method
33mWrangling Internet-scale Image Datasets
01h 30mAdvanced Software Testing for Data Scientists