Double normal distribution

author: Thomas van Tongeren

date: 22 nov 2014

Normal distributions

Gaussian/Normal distributions

The normal distribution is well known since it is the result of many independent processes. The normal distribution describes a distribution of numbers with the middle and the spread of a distribution of values. However if one fits it to the faitful eruptions dataset one can reasonably see the mean is not a meaningful result, since almost no geiser has an eruption time of 3 seconds.

Course Project

In this shiny app we show the user something about a different distributions, without the requirement that the student learns a new one. This could be useful from an educational point of view.

Problem definition: distribution is not representative

plot of chunk unnamed-chunk-1

[1] 3.488

Solution to the problem: make a superposition of gaussians

There are not 2 values (sd and mean), but four values, 2 for each Gaussian.

plot of chunk unnamed-chunk-2

Outlook: Fit the data to more than 2 Gaussians.

Ofcourse the next step is to automatically fit the distribution to the sample. In the case of an automatically fitting algorithm, one can also include more and more Gaussians to fit more complex distributions.