# How to fit a Gaussian distribution with outlier data points?

I have a data set which consists of large number of data points. However, there are some outlier points that can be considered be noise. If I include all data points to approximate the Gaussian distribution, the standard deviation is apparently larger than expected. How can I fit this data set with a Gaussian distribution and get accurate mean and standard deviation ignoring these noise points.

Thanks

• is it a univariate or a multivariate gaussian you are trying to fit? – user603 Nov 11 '13 at 15:12
• a univariate gaussian, not mixture gaussian – user22062 Nov 12 '13 at 5:27
• multivariate is not the same as mixture. Assuming you meant univariate, then, your question is a duplicate of this one – user603 Nov 12 '13 at 9:15
• The question to ask yourself: how do you know, that they are outliers? – Tim Jan 3 '17 at 12:44

## 1 Answer

Try https//mycurvefit.com. It's an online tool that will fit a curve to your data. (AFAIK its free for smaller data sets but there might be a charge for large data sets). With the tools just enter your data, select the fit. You can also mark points as outliers by clicking to "flag" then remove those outliers and recalculate the new curve, e.g. • -1 This has nothing to do with fitting distribution to data. Fitting distribution to data $\ne$ fitting curve to points. – Tim Feb 4 '17 at 17:14