Keep it Clean - Why Bad Data Ruins Projects and How to Fix It

Explore the critical impact of data quality on machine learning projects in this 44-minute conference talk from GOTO Chicago 2019. Discover why bad data is costing the US economy trillions and learn practical techniques to identify and fix data issues. Delve into high-profile case studies, including Microsoft's Tay.ai chatbot, to understand the consequences of poor data management. Gain insights into the Compass Model, AI in healthcare, and challenges in Google Translate. Examine common data problems such as noise, biases, and leakage, and master strategies for data visualization, consistency, and anomaly detection. Apply these concepts to real-world examples like the Titanic dataset and wine quality analysis. Equip yourself with essential skills to improve data quality and boost the success of your machine learning projects.

Intro
The Compass Model
Babylon Health
AI Doctor
News Removed
Google Translate
Adverse adversarial attacks
The numbers
Costing money
Why this happens
Translation
Noise
Biases
Visualization
Data Availability
Data Consistency
Data Leakage
Titanic Dataset
How to Fix Missing Data
Feature Noise
Noise Reduction
Anomalies
Wine Quality