Link to home
Start Free TrialLog in
Avatar of Saqib Husain
Saqib HusainFlag for Pakistan

asked on

VBA: Getting a random number

I am trying to get a random number with 4 significant figures using this formula

num = Round(Rnd(), 4) * 10 ^ Int((Rnd * 10) - 5)

Although the components work out nicely the final result comes up with lesser significant figures. How can I get only 4 significant figures in VBA?
Avatar of zorvek (Kevin Jones)
zorvek (Kevin Jones)
Flag of United States of America image

num = Round(Rnd() * 10 ^ Int((Rnd * 10) - 5), 4)

Kevin
What is the range of values you are seeking? Maximum and minimum?

Kevin
Avatar of Saqib Husain

ASKER

10^ (+/-5)
Does this work?

Round(Rnd() * 10 ^ Int(Rnd() * 10 - 5), 4)

Kevin
Something weird is going on. There appears to be a bug in the VBA interpreter.

This:

Round(Rnd() * 10000, 4)

does not round to four places. It doesn't seem to round at all.

This works:

Dim Value As Double
Value = Rnd() * 10000
Value = Round(Value, 4)

Kevin
Currently I have applied the same formula to an excel cell and used the value of the cell. But my question still stands....how to get it directly through VBA?
I showed you how above. That's all VBA dude.

Kevin
This does not give me the desired results

pwr = Rnd()
pwr = 10 ^ Int((pwr * 10) - 5)
num = Rnd()
num = Round(num, 4)
num = num * pwr
Convert to currency data type, which has four decimal places.
Example:
num = CCur(Rnd() * 10 ^ Int((Rnd * 10) - 5))

Open in new window

aikimark, I am looking for numbers in the range

0.00001000 to 999900

and I do not think the currency will do this as 0.00001001 is a possibility.
so are you looking for something like this?

cdbl(rnd()*1000000)

sample result:  289562.463760376
ASKER CERTIFIED SOLUTION
Avatar of andrewssd3
andrewssd3
Flag of United Kingdom of Great Britain and Northern Ireland image

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
@ssaqibh

You stated that you wanted a "number with 4 significant figures".  That generally means that you want a floating point number with a precision of four digits to the right of the decimal point.

Your two recent examples
0.00001000 and 0.00001001 go beyond your stated precision.  If displaying such values with a precision of four decimal places, both of these would be zero.
Not sure what Excel has built-in, but in Access VBA, I wrote the code below to round to a significant number of digits.  Probably better ways of doing this, but it did achieve the correct results.

Just feed it a random number and I believe it will give you what you want.

Jim.

Public Function RoundSignificant(varValue As Variant, intNumSignificantDigits As Integer) As String

    Dim strPrefix As String
   
    Dim dblFactor As Double
    Dim dblABSofValue As Double
    Dim strFormatedValue As String
   
    Dim lngPos As Long
    Dim bolDecimalPoint As Boolean
    Dim strChr As String
   
    Dim lngNumOfDigits As Long
       
   
    ' Check for a prefix ('>' or '<').  If there is one,
    ' strip it off for now.
    If left(varValue, 1) = ">" Or left(varValue, 1) < "<" Then
      strPrefix = left(varValue, 1)
      varValue = Val(Mid(varValue, 2))
    Else
      strPrefix = ""
    End If
   
    ' Check for null
    varValue = Nz(varValue, 0)

    ' If value of zero, return "N/A"
    If varValue = 0 Then
      RoundSignificant = "N/A"
    Else
     
      ' Get the factor
      dblFactor = 10 ^ Int(Log(Abs(varValue)) / Log(10) - intNumSignificantDigits + 1)
     
      ' Based on the factor, get an absolute value that's rounded.
      If dblFactor = 1 Then
         dblABSofValue = CLng((Abs(varValue) / dblFactor))
      ElseIf dblFactor > 1 Then
        dblABSofValue = CLng((Abs(varValue) / dblFactor) + (0.5 / dblFactor))
      Else
        dblABSofValue = CLng((Abs(varValue) / dblFactor) + (0.5 * dblFactor))
      End If
      dblABSofValue = dblABSofValue * dblFactor
      ' Format the value as a string.
      strFormatedValue = Format((IIf(varValue >= 0, 1, -1) * dblABSofValue), "#0.00000000000000000000")
     
      ' Do we have a decimal point?
      If InStr(strFormatedValue, ".") > 0 Then
        ' If so, chop off all zeros on the right
        While right(strFormatedValue, 1) = "0"
            strFormatedValue = left(strFormatedValue, Len(strFormatedValue) - 1)
        Wend
      End If
     
      ' Scan for the number of digits in the string
      For lngPos = 1 To Len(strFormatedValue)
        strChr = Mid(strFormatedValue, lngPos, 1)
           
        If strChr >= "0" And strChr <= "9" Then
          lngNumOfDigits = lngNumOfDigits + 1
        End If
       
        If strChr = "." Then
          bolDecimalPoint = True
        End If
      Next
     
      ' Is the number of digits found less then the significance required?
      If lngNumOfDigits < intNumSignificantDigits Then
        ' If so and we have decimal point, add some zeros
        If bolDecimalPoint = True Then
          strFormatedValue = strFormatedValue & String(intNumSignificantDigits - lngNumOfDigits, "0")
        End If
      End If
         
      ' Do we have anything to the right of the decimal?
      ' If not, remove the decimal point
      If right(strFormatedValue, 1) = "." Then
        strFormatedValue = left(strFormatedValue, Len(strFormatedValue) - 1)
      End If
     
      ' Add the prefix back in if we have one
      If strPrefix <> "" Then
        RoundSignificant = strPrefix & strFormatedValue
      Else
        RoundSignificant = strFormatedValue
      End If

    End If

End Function
rspahitz
sample result:  289562.463760376
Nope Only 4 significant figures



andrewssd3
I still have to test your solution

aikimark
You stated that you wanted a "number with 4 significant figures".  That generally means that you want a floating point number with a precision of four digits to the right of the decimal point.
examples of 4 significant figures are
1234000000
1234000
12340
123.4
12.34
1.234
0.1234
0.001234
0.000001234



 JDettman
I am too scared to try your solution unless you confirm that it meets my requirements as per the list ending four lines above
FYI, just found the formula for significant rounding:

round(10-n·x)·10n, where n = floor(log10 x) + 1 - p

x being a positive number to a precision of p significant digits.  I believe this will work in Excel.  For negatives use the abs() value.

Jim.
oops, no super scripts.  That's the round of 10 raised to the negative n times X times 10 rasied to n.

 Not sure if that will work or not, but you can try it.

Jim.
<<I am too scared to try your solution unless you confirm that it meets my requirements as per the list ending four lines above >>

 Just cut and paste into a module and then call it from a cell, but the formula I just posted is a lot more conscise for Excel.

Jim.
This will give you a four digit integer.
Int(Rnd() * 1000) 

Open in new window


It is up to you to multiply it by some power of 10 that will move the decimal place to the right or left.  In your example, you wanted the power raised (decimal moved) from +2 (right) to -8 (left).
Num = Int(Rnd() * 1000) * 10^(Int(Rnd() *11)-8)

Open in new window


Note: your Num variable should be a Double data type.
Hi - I have just tested my solution in 100,000 iterations, and it gives

Max - 999900
Min - 0.00001

Which I think is what you wanted.
So you want 4 digits, but randomly placed within powers of 10?
Int(Rnd()*10000) will produce a number from 0 to 9999 (although some times it may round to 10000 so you may want a separate check for that).

then just multiple or divide by powers of 10.

So using your original example,

num= Int(Rnd()*10000) * 10 ^ Int((Rnd * 10) - 5)

However, sometimes you might get random numbers like 0123 so does that count as only 3 significant digits?  If so you really want a range of 1000-9999 so here:
num= Int(Rnd()*9000+1000) * 10 ^ Int((Rnd * 16) - 8)

Open in new window

Sample data from the above:

 0.004129
 93110
 0.0009643
 67.09
 0.0005886
 62970000
 285400000
 140.4
 3.291
 311.7

I just noticed that sometimes you will get less than 4 digits...that's because after multiplying by 9000, the right-most digit might be zero.  I guess you can prevent that by ensuring that your resulting number after adding 1000 is not divisible by 10.  the easiest way to handle that is to simply generate a new number since 90% of the time you'll be okay.

Do
tempNum=Int(Rnd()*9000+1000)
Loop Until (tempNum Mod 10)<>0
num= tempNum * 10 ^ Int((Rnd * 16) - 8)

Open in new window

If it helps, here's what I used for testing:

    Dim num As Double
    Dim origNum As Double
    Dim tempNum As Double
    Dim iCntr As Integer
    
    For iCntr = 1 To 100
        Do
            origNum = Rnd
            tempNum = Int(origNum * 9000 + 1000)
            'If tempNum Mod 10 = 0 Then
            '    Debug.Print ;
            'End If
        Loop Until (tempNum Mod 10) <> 0
        num = tempNum * 10 ^ Int((Rnd * 16) - 8)
        Debug.Print origNum; tempNum; num 'Int(num * 9000 + 1000) * 10 ^ Int((Rnd * 16) - 8)
    Next

Open in new window


The Immediate window will show the original number, the number before the power (but after multiplying by 9000 and adding 1000, then the final number.

Sample results:

 0.775829255580902  7982  798200000
 0.893206655979156  9038  903.8
 0.255511462688446  3299  3.299
 0.769428551197052  7924  792400000
 0.173246800899506  2559  25.59
 0.314691126346588  3832  0.003832
 0.651410758495331  6862  0.00006862

So this will produce a 4-digit number where neither the leading nor trailing digit is zero, since apparently that is not considered significant.  i.e. 1230 is no good because it only has 3 significant digits.  Likewise for 0.01230 but okay for 1001.
First solution that passes.

Thanks
Good idea, but it will fail on numbers that are powers of 10, like 9370, which might appear as 9370000 or .00937
It will also produce numbers from 1 to 999, which will leave you without those "4 significant digits"
rspahitz, those numbers are part of the range of 4 significant figures. In one of my clarifications I specified the range

0.00001000 to 999900

which indicates that 0.00001000 is a member.
not to argue, but then that means that 0.00001 is valid, since it's the same as 0.00001000 (but formatted differently.)
But glad you got your working answer.
No problem with getting things clarified.

Yes it is valid.
My original question says

How can I get only 4 significant figures in VBA?


Which probably should have been

How can I get not more than 4 significant figures in VBA?

to eliminate the confusion.