Floating point pitfalls

John D. Cook over at the Endeavour has a series of articles talking about floating-point arithmetic and how it can burn us in computing statistics like the standard deviation, correlation and regression coefficients using the book formulae. Specially enlightening for me was the trick of using the Taylor series expansion of log(1+x) for small values of x, since the error is actually quite small. Fantastic points, John!!

A good summary of his points can be found here

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s