Find solution
German trade statistics?

Besides statistical significance/practical significance trade-off, bias/variance trade-off, what are other important trade-offs in Statistics and Machine Learning?

  • Better provide interesting explanations and examples.

  • Answer:

    Type I vs Type II error.  Are you willing to accept more false positives or false negatives? Estimation vs. prediction - is your primary interest in the estimation of parameters (the effect size of variables) or in the prediction of responses? Simple vs. complex. These are actually all related and equivalent, mathematically.  That includes significance and bias-variance. Powerful tests (high Type I error rates, low Type II error rates) tend to detect practical significance.  Predictive models are often overparameterized black box models with low bias, but high variance.  These overparameterized models are often built with zero significance thresholds, thus favoring power.

Justin Ma at Quora Visit the source

Was this solution helpful to you?

Other answers

Statistical significance vs. practical importance is not a trade-off. You do not increase one when you decrease the other (at least, not typically). Other trade-offs: Approximate solution to the right problem vs. an exact solution to the wrong problem. Using a method that is proper for the problem at hand vs. one that is understandable to the audience/will get published.

Peter Flom

An example of statistical significance / practical significance trade-off: If a higher confidence level means that we are more confident about the number we are reporting, why don’t we always report a confidence interval with the highest possible confidence level? Higher CI level is accompanied with larger CI range, and sometimes comparatively extreme values and large range have no practical meaning. For example, what's the point of saying "We are 100% confident that human beings' average height is within 0 and 2 meters" ?

Shuai Wang

Time and Perfection. In Kaggle and predictive modeling competitions, you could invest a few weeks to squeeze that last 1% from our model. In data science applications, you would probably forgo the additional 1% in order to start a new project with potentially higher impact.

William Chen

Related Q & A:

Just Added Q & A:

Find solution

For every problem there is a solution! Proved by Solucija.

  • Got an issue and looking for advice?

  • Ask Solucija to search every corner of the Web for help.

  • Get workable solutions and helpful tips in a moment.

Just ask Solucija about an issue you face and immediately get a list of ready solutions, answers and tips from other Internet users. We always provide the most suitable and complete answer to your question at the top, along with a few good alternatives below.