A naukri.com initiative
Towards Data Science
4w
434
The Dangers of Deceptive Data Part 2–Base Proportions and Bad Statistics
- The article is a follow-up to a previous piece on deceptive data and explores statistical concepts leading to data misinterpretation.
- It emphasizes the distinction between correlation and causation, highlighting the need for controlled trials to establish causation.
- An example involving cigarette consumption and life expectancy illustrates the correlation vs. causation fallacy.
- The base rate fallacy is discussed using an example from a medical school scenario to demonstrate the importance of base proportions in statistics.
- The article mentions the Datasaurus Dozen dataset, showing how summary statistics alone may not reveal the full picture of the data.
- It stresses the significance of understanding uncertainty and explains how error bars represent uncertainty, not errors in data.
- The article cautions against mistrusting correct data due to misinterpretation of uncertainty and the societal implications of such misunderstandings.
- Key takeaways include being cautious of deceptive data, considering base rates, exploring data beyond summary statistics, and interpreting uncertainty accurately.
- Understanding these concepts equips individuals to approach data science problems effectively and make informed decisions.
- The article concludes by encouraging readers to stay vigilant against deceptive data and misinformation.
Read Full Article
26 Likes
For uninterrupted reading, download the app