Conditional sum (i.e.: sum if) over a vector
Posted: Mon Jun 15, 2015 3:12 am
Dear Members,
I am having a presumably very common problem:
Simulating a time series many times, then conducting a test, and then check how often the Null hypothesis was rejected.
That way I can test the size and power of the test. At the end I store the p-values in a vector (not a series).
I need the sum of entries in that vector that are less then 5%.
And I dont find a command for this. Like the simple Excel "sumif".
The question might have been answered a million times, but I didnt find any of these.
The Problem can be solved by a for loop such as:
Question: Is there a command for this?
Maybe I can use @sumsby. But then it seems to me that I have to (a) use a series instead of a vector which is not so nice and (b) first have to specify a new series containing a 1 if pvalue<0.05 and a zero otherwise. Doing (b) would again require a loop such as the one above, doesn't it?
To add some naivety to my question: How can I, generally, find commands like the sumsby? Or find a command that does what I am looking for in the above question? I often only get anywhere by searching the forum and I am having a hard time finding what I am looking for in the Help files. Am I doing somewhing wrong?
Thanks 2 everyone!
Daniel
I am having a presumably very common problem:
Simulating a time series many times, then conducting a test, and then check how often the Null hypothesis was rejected.
That way I can test the size and power of the test. At the end I store the p-values in a vector (not a series).
I need the sum of entries in that vector that are less then 5%.
And I dont find a command for this. Like the simple Excel "sumif".
The question might have been answered a million times, but I didnt find any of these.
The Problem can be solved by a for loop such as:
Code: Select all
count=0
for !i=1 to !mcmax
if pvalue(!i) <0.05 then
count= count+1
endif
next
Question: Is there a command for this?
Maybe I can use @sumsby. But then it seems to me that I have to (a) use a series instead of a vector which is not so nice and (b) first have to specify a new series containing a 1 if pvalue<0.05 and a zero otherwise. Doing (b) would again require a loop such as the one above, doesn't it?
To add some naivety to my question: How can I, generally, find commands like the sumsby? Or find a command that does what I am looking for in the above question? I often only get anywhere by searching the forum and I am having a hard time finding what I am looking for in the Help files. Am I doing somewhing wrong?
Thanks 2 everyone!
Daniel
scalar count = @sum(@elt(pwerte_05,0.05*@ones(!mcmax)))