How to choose MaxNumChanges when using findchangepts

3 次查看（过去 30 天）

Gianluca Finotti 2018-4-27

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/397819-how-to-choose-maxnumchanges-when-using-findchangepts

评论： Greg Dionne 2018-6-4

Hi, I am using findchangepts to find the points where a signal changes the most. This is not really an electric signal. These are participant's responses through time using a potentiometer. So, on a table I have a linear slide potentiometer like this:

Participants are using this to rate whether a particular feeling during the experiment is increasing or decreasing.

I am using findchangepts to understand when there are significant changes in their response. I arbitrarily choose MaxNumChanges, 4. This is the line of code:

findchangepts(myVar,'Statistic','linear','MaxNumChanges',4)

The result seems quite good. Given what I expected, it seems to make sense:

However, the error seems to be huge. Also, I don't know if I can show this in a publication. I expect reviewers to ask me why did I choose 4 points and not 2 or 10, and also why I have such a huge error.

So, does anybody know how can choose the MaxNumChanges and what would be an acceptable error?

Thank you for your time, Gianluca