10 ms·

Might be a little premature to call for retirement of sigma. The mathematical concept of standard deviation is super useful but I agree that the name is confusi

Might be a little premature to call for retirement of sigma. The mathematical concept of standard deviation is super useful but I agree that the name is confusing and that we need to improve the naming and ideally the notation and teaching of statistics. Ability to deal with uncertainty and variance is becoming more and more important in all sorts of fields as data volumes get larger so I'd hate to see us give up just because it is hard to understand.

HelloNurse 10y ago
"Dealing with variance" is the same as dealing with mean average deviation. Variance is "super useful" when one works with Gaussian distributions, and overusing variance is part of the greater problem of overusing Gaussian distributions and Euclidean distances.
CorvusCrypto 10y ago
agree. You have to wonder if these articles come up as a way to draw people in to looking back at the theory to justify their use. A sort of "trolling" to get people back on board thinking of the theory and not just the use. Both forms of deviation measurement have their use, along with the other many forms of deviation measurement used in stats.
stuxnet79 10y ago
I've never really been a fan of the notation. But I don't know how you can conceivably enforce this because just like programming, people will pontificate about 'design', 'clean code', 'maintanability' and other similar cargo-cult buzzwords but they will go ahead and do whatever they want with their code.
- chuckcode 10y ago
  Completely agree that standardizing notation takes decades as a lot of it is how textbooks and tutorials from experts are written. I for one though am really glad that we standardized on Leibniz's notation rather than Newton's for calculus though and aren't using roman numerals anymore.
xapata 10y ago
Squared curves have a nice-looking derivative. Absolute value curves have a constant derivative. The only reason the squared version won was to make the calculus easy. Now that we have computers, the absolute value curve is better for most uses.

HelloNurse 10y ago

"Dealing with variance" is the same as dealing with mean average deviation. Variance is "super useful" when one works with Gaussian distributions, and overusing variance is part of the greater problem of overusing Gaussian distributions and Euclidean distances.

CorvusCrypto 10y ago

agree. You have to wonder if these articles come up as a way to draw people in to looking back at the theory to justify their use. A sort of "trolling" to get people back on board thinking of the theory and not just the use. Both forms of deviation measurement have their use, along with the other many forms of deviation measurement used in stats.

stuxnet79 10y ago

I've never really been a fan of the notation. But I don't know how you can conceivably enforce this because just like programming, people will pontificate about 'design', 'clean code', 'maintanability' and other similar cargo-cult buzzwords but they will go ahead and do whatever they want with their code.

chuckcode 10y ago
Completely agree that standardizing notation takes decades as a lot of it is how textbooks and tutorials from experts are written. I for one though am really glad that we standardized on Leibniz's notation rather than Newton's for calculus though and aren't using roman numerals anymore.

xapata 10y ago

Squared curves have a nice-looking derivative. Absolute value curves have a constant derivative. The only reason the squared version won was to make the calculus easy. Now that we have computers, the absolute value curve is better for most uses.