
Talk Python To Me #497: Outlier Detection with Python
67 snips
Mar 21, 2025 Brett Kennedy, a seasoned software developer with 30 years of expertise in data science, shares his insights on outlier detection. He explains how outliers can uncover fraud or hidden patterns in data, emphasizing their critical role in various fields, including finance and biology. The discussion covers Python's unique methods for outlier detection, the logistical challenges of massive datasets, and the importance of ongoing model retraining. Brett also explores advanced techniques and tools, stressing the balance between dataset size and computational resources.
AI Snips
Chapters
Transcript
Episode notes
Pulsars and Outliers
- Pulsars were discovered through manual analysis of astronomical data.
- Now, outlier detection helps find interesting phenomena in petabytes of nightly data.
Mailing Hard Drives
- Brett Kennedy and Michael Kennedy discussed mailing hard drives of financial data.
- This was due to prohibitive internet costs 15 years ago.
Outliers and New Science
- Outliers can reveal unexpected information, leading to new scientific discoveries.
- An example is in astronomy where unusually bright spots could signify a quasar.

