Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have this problem where I have a set of data. Each data point has two values:

Value1: Birthday
Value2: Height

Now, the problem I am faced with involves finding the optimal groupings of data points to minimize the percent variance of Birthdays and Heights in each group (there are other additional constraints that I won't go into here). Finding the percent variance of Height is very straight forward, but for Birthday, it is more complex.

To calculate the percent variance of Birthday, my strategy is to count the numbers after the date January 1st, 0 AD of each Birthday, calculate the average, and average the squared percent difference of each Birthday to get the percent variance. That's all well and good, but because the number of days after January 1st, 0 AD is so huge, the problem I'm having is that the percent variance is tiny compare to the percent variance in height.

My strategy is to use percent variance so I can normalize the variances in Birthday and Height so I can measure the quality of a grouping by comparing these variances, but because the Birthday percent variance is so small, it's hard to determine when a real improvement is made in groupings based on Birthday variance.

Are there any other ways I can calculate a percent variance for Birthday, so that it is more comparable to the variance in Height? Are there any strategies that I could use to help improve my results?

share|cite|improve this question
as an aside: there is no such year in the western calendar as 0AD, 1AD is immediately preceded by 1BC. – Willie Wong Jun 30 '11 at 14:12
Have you tried measuring birthdays from the mean of all the birthdays? – Jay Jun 30 '11 at 14:19
@Willie, interesting fact, thanks – Soo Jun 30 '11 at 14:25
@Jay, in percentage terms, this still has the same problems as just measuring birthdays from the mean of the birthdays in a particular group. – Soo Jun 30 '11 at 14:26
up vote 1 down vote accepted

What if you first scale all birthdays and heights so that they fall between 0 and 1? E.g., if all the heights are between, say, 60 inches and 75 inches, then replace each height $h$ with $(h-60)/15$?

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.