# Standard Deviation question regarding EDOM errors

**URL:** <https://www.ruby-forum.com/t/standard-deviation-question-regarding-edom-errors/172469>\
**Category:** Ruby\
**Created:** [July 22, 2009, 12:32am UTC](https://www.ruby-forum.com/t/standard-deviation-question-regarding-edom-errors/172469 "2009-07-22T00:32:42Z")\
**Posts on this page:** 9\
**Page:** 1

<div class="post-metadata">

**Author:** ![elricstorm](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/e/da6949/32.png) [@elricstorm](https://www.ruby-forum.com/u/elricstorm)\
**Post date:** [July 22, 2009, 12:32am UTC](https://www.ruby-forum.com/t/standard-deviation-question-regarding-edom-errors/172469/1 "2009-07-22T00:32:42Z")

</div>

Okay,

Take a very close look at the two (one is a class, the other is a  
method):

class Stdev  
def initialize(arr)  
@arr = arr  
@size = @arr.to\_a.size  
end

def sum  
@arr.inject {|a,b| a + b }  
end

def product  
@arr.inject {|a,b| a \* b }  
end

def calculate  
Math.sqrt(((product - (sum \* sum)/@size)/(@size-1)))  
end

end

def stddev(arr)  
size = arr.to\_a.size  
product, sum = 0.0, 0.0  
arr.each do |a|  
sum += a  
product += a\*a  
end  
std = Math.sqrt(((product - (sum \* sum)/size)/(size-1)))  
return std  
end

Testing with both:

a = [123.369, 68.6179]

Stdev.new(a).calculate  
=\> Errno::EDOM: Domain error - sqrt

stddev(a)  
=\> 38.7148740874228

Test a web calculator for the same thing:

[http://www.easycalculation.com/statistics/standard-deviation.php](http://www.easycalculation.com/statistics/standard-deviation.php)

The method will pan out.

So, to save suspense the reason why the class is erroring out is because  
the sqrt is being attempted on a negative number. Therefore, in order  
to continue to use that I would have to require ‘complex’ and the return  
values from complex return two values in an array.

I’m just curious, why, given both of these being so very similar - the  
method passes but the class fails?

---

<div class="post-metadata">

**Author:** ![elricstorm](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/e/da6949/32.png) [@elricstorm](https://www.ruby-forum.com/u/elricstorm)\
**Post date:** [July 22, 2009, 12:42am UTC](https://www.ruby-forum.com/t/standard-deviation-question-regarding-edom-errors/172469/2 "2009-07-22T00:42:28Z")

</div>

Hi –

On Wed, 22 Jul 2009, Älphä Blüë wrote:

> def product  
> @arr.inject {|a,b| a \* b }  
> end

> product, sum = 0.0, 0.0  
> arr.each do |a|  
> sum += a  
> product += a\*a  
> end

The logic in the second snippet is wrong (assuming the logic in the  
product method is what you want). Try it with [1,2], for example, and  
you’ll get 2 for the first one and 5 for the second.

David

---

<div class="post-metadata">

**Author:** ![elricstorm](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/e/da6949/32.png) [@elricstorm](https://www.ruby-forum.com/u/elricstorm)\
**Post date:** [July 22, 2009, 12:51am UTC](https://www.ruby-forum.com/t/standard-deviation-question-regarding-edom-errors/172469/3 "2009-07-22T00:51:16Z")

</div>

> The logic in the second snippet is wrong (assuming the logic in the  
> product method is what you want). Try it with [1,2], for example, and  
> you’ll get 2 for the first one and 5 for the second.
> 
> David

Hi David,

Please bear with me as my brain is a bit fried - been working on a lot  
of complex issues all day long and so my mind is mush at this point. I  
don’t quite follow…

The method produces correct standard deviation results everytime…

a = [1,2,3,4,5]  
b = [1,2]  
c = [123.369, 68.6179]

stddev(a)  
=\> 1.58113883008419

stddev(b)  
=\> 0.707106781186548

stddev©  
=\> 38.7148740874228

---

<div class="post-metadata">

**Author:** ![elricstorm](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/e/da6949/32.png) [@elricstorm](https://www.ruby-forum.com/u/elricstorm)\
**Post date:** [July 22, 2009, 12:53am UTC](https://www.ruby-forum.com/t/standard-deviation-question-regarding-edom-errors/172469/4 "2009-07-22T00:53:49Z")

</div>

Hi David,

I think I understand what you are saying…

What I want with…

product += a\*a

is … a[0]\*a[0]+a[1]\*a[1]+a[2]\*a[2]…etc.

---

<div class="post-metadata">

**Author:** ![elricstorm](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/e/da6949/32.png) [@elricstorm](https://www.ruby-forum.com/u/elricstorm)\
**Post date:** [July 22, 2009, 1:04am UTC](https://www.ruby-forum.com/t/standard-deviation-question-regarding-edom-errors/172469/5 "2009-07-22T01:04:41Z")

</div>

Hi –

On Wed, 22 Jul 2009, Älphä Blüë wrote:

> of complex issues all day long and so my mind is mush at this point. I  
> don’t quite follow…

The method is calculating “product” as the sum of the squares (which I  
don’t actually think is the product, is it?). The class’s product  
method is calculating the product of all the elements of the array. So  
they’re different.

[1,2,3].inject {|a,b| a \* b } # 6

total = 0.0  
[1,2,3].each {|a| total += a \* a }  
total # 14.0

David

---

<div class="post-metadata">

**Author:** ![elricstorm](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/e/da6949/32.png) [@elricstorm](https://www.ruby-forum.com/u/elricstorm)\
**Post date:** [July 22, 2009, 1:05am UTC](https://www.ruby-forum.com/t/standard-deviation-question-regarding-edom-errors/172469/6 "2009-07-22T01:05:16Z")

</div>

Hi –

On Wed, 22 Jul 2009, Älphä Blüë wrote:

> Hi David,
> 
> I think I understand what you are saying…
> 
> What I want with…
> 
> product += a\*a
> 
> is … a[0]\*a[0]+a[1]\*a[1]+a[2]\*a[2]…etc.

OK, then the product method in the class is wrong. “Product” doesn’t  
sound like the right name for a sum-of-the-squares calculation. But  
anyway –

def whatever\_you\_call\_it # 🙂  
@arr.inject(0) {|acc,x| acc + x \* x }  
end

David

---

<div class="post-metadata">

**Author:** ![elricstorm](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/e/da6949/32.png) [@elricstorm](https://www.ruby-forum.com/u/elricstorm)\
**Post date:** [July 22, 2009, 1:21am UTC](https://www.ruby-forum.com/t/standard-deviation-question-regarding-edom-errors/172469/7 "2009-07-22T01:21:31Z")

</div>

Hi David,

Thanks for showing me my error. So, I implemented a change in the class  
for testing purposes:

class Stdev  
def initialize(arr)  
@arr = arr  
@size = @arr.to\_a.size  
end

```
def sum
  @arr.inject {|a,b| a + b }
end

def product
  total = 0.0
  @arr.each {|a| total += a * a }
  total
end

def calculate
  Math.sqrt(((product - (sum * sum)/@size)/(@size-1)))
end

```

end

Running the same test with the same things:

a = [1,2,3,4,5]  
b = [1,2]  
c = [123.369, 68.6179]

stddev(a)  
=\> 1.58113883008419 (correct)

stddev(b)  
=\> 1.0 (NOT expected)

stddev©  
=\> 38.7148740874228 (correct)

So, compared to the class now, versus the method I created, do you see  
any anomolies that would change this behavior?

---

<div class="post-metadata">

**Author:** ![elricstorm](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/e/da6949/32.png) [@elricstorm](https://www.ruby-forum.com/u/elricstorm)\
**Post date:** [July 22, 2009, 2:30am UTC](https://www.ruby-forum.com/t/standard-deviation-question-regarding-edom-errors/172469/8 "2009-07-22T02:30:54Z")

</div>

Thanks so much David - that’s perfect…

Works great now as a class…

---

<div class="post-metadata">

**Author:** ![elricstorm](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/e/da6949/32.png) [@elricstorm](https://www.ruby-forum.com/u/elricstorm)\
**Post date:** [July 22, 2009, 2:27am UTC](https://www.ruby-forum.com/t/standard-deviation-question-regarding-edom-errors/172469/9 "2009-07-22T02:27:05Z")

</div>

Hi –

On Wed, 22 Jul 2009, Älphä Blüë wrote:

> def sum  
> @arr.inject {|a,b| a + b }  
> end

@arr.inject(0.0) {|a,b| a + b }

which will give you a float error (which corresponds to what you have  
in the method version).

> So, compared to the class now, versus the method I created, do you see  
> any anomolies that would change this behavior?

See above 🙂

David