Archives September 2019

The Bias/Variance Trade-off And The Limits of Machine Learning Models

Last update on Sept. 30, 2019.

Image Credit: pathdoc / Shutterstock.com

 

A Machine Learning model is a bit like a marriage - no matter how hard you try, it's never going to be perfect. The same way every marriage has its issues, so does every machine learning model. Real world data is like real world people: unpredictable. Real world events don’t conform precisely to a perfectly predictable pattern, so no matter how clever your algorithm, there will always be some irreducible error. The trick is to focus on the things you can change - the reducible error. If you’re in a marriage where there ...

Continue reading

Is Passive Investing A Bubble?

Last update on Sept. 16, 2019.

Image Credit: EtiAmmos / Shutterstock.com

 

Within each industry, there are certain topics that divide participants and make them battle each other with religious fervour. Among programmers, the topic might be Windows vs. *nix. Among physicists, I’ve heard of dissent against string theory.

There are several topics that similarly divide financial professionals. One such topic revolves around the issue of whether passive investing is harmful or beneficial to financial markets. Passive investing, in contrast to active investing, involves an attempt not to pick individual stocks. Instead, it diversifies as much as possible, owning every stock in an ...

Continue reading

Outlier Detection: Border Security For Your Model

Last update on Sept. 2, 2019.

Image Credit:  zhu difeng / Shutterstock.com

 

You might not know it, but outlier detection is a big part of your life. The ability to notice  diseased or rotten food, or spot a person in a crowd who might be unstable or dangerous is a very important human skill. Many of us had extra detection-training as children

So can we reliably spot outliers? For extreme examples, we're pretty good. Say you ask 100 people, “what is the ideal indoor temperature?” and 99 of them answer, “somewhere between 16 and 22 degrees Celsius”, while the other ...

Continue reading



Web Analytics