Sumproduct issues

Hi Harlan,



I want to thank you for being detailed and taking the time to explain SUMPRODUCT, data types, coercion etc. I learned a lot reading your posts in this thread. It has been educational. I'll take the following advice to heart.



--x is guaranteed to convert text representations of numbers to their numeric values (with possible floating point rounding/truncation error) while leaving numeric values unchanged, and x&"" is guaranteed to convert numeric values to their text representations while leaving text as-is.



In my opinion, the fact that you are meticulous/precise on terminology is a reflection of your expertise. I always believe that programmers do well in Excel because writing a formula is like writing a line of code in a program. A sense of perfection is an important quality of a good programmer.



I also want to thank you for explaining #N/A (wildcard) to me the other day.



Last but not least, I don't think you have insulted anybody.



I am going to go over your posts again and if I have any questions, I'll start my own thread. If I do, I hope you have time to drop by.



We are lucky to have you around sharing your expertise and *generosity*. Please don't give up on us.



I hope you still come to this thread and can read this.



Epinn


"Harlan Grove" wrote in message oups.com...
SteveDB1 wrote...
Ok, Harlan,
I tried the last two equations you mentioned, and the first one does not
work. It only returns a zero.
(--Map!$E$4:$E$30=$A3)

or

(Map!$E$4:$E$30=$A3&"")

Simply put, I don't believe you. If you can get the result you claim to
want using formulas like =1*Map!E4 in a different range of 27 rows by 1
column and using that range rather than Map!E4:E30 in your SUMPRODUCT
formula, then Excel would give the same result using --Map!E4:E30 in
the SUMPRODUCT formula. I've NEVER seen an exception to that, and I'm
going to be skeptical about claims otherwise until I see a live
example.

Maybe you meant that Map!E4:E30 is numeric but A3 isn't. In that case
use

(Map!$E$4:$E$30=--$A3)

or

(Map!$E$4:$E$30&""=$A3)

But to be as generic as possible, use

(-Map!$E$4:$E$30=-$A3)

or

(Map!$E$4:$E$30&""=$A3&"")

So, all of this still brings me back to the point of-- how do I,
consistently, constantly, get it to work, without having to go through and
find out why one equation works and the very next one does not work.

There's NOTHING wrong with Excel's -- double operator or &"" or
SUMPRODUCT. The problem lies entirely in probable data type mismatches
between your ranges Map!E4:E30 and A3. It's entirely and exclusively up
to YOU to ensure that you're comparing text to text or numbers to
numbers.

My whole point for coming in here to ask is that the equation is not
consistently working, and I'm spending far too much time going through
looking for small, inconspicuous reasons for its failure. I came looking to
learn more about this equation because it did once work, and it sped up my
work process by an undefinable amount of time. It's now taking 3 x's longer
to figure out why it's not working.

It's not an 'equation'. It's a formula.

If your formulas appear to behave inconsistently, far & away the most
likely answer is that there are problems with YOUR data. To repeat,
it's up to YOU to ensure that you're comparing text to text or numbers
to numbers. There are ways to do that: --x is guaranteed to convert
text representations of numbers to their numeric values (with possible
floating point rounding/truncation error) while leaving numeric values
unchanged, and x&"" is guaranteed to convert numeric values to their
text representations while leaving text as-is. You could also have
trailing nonbreaking HTML spaces, and Roger Govier has already shown
how to handle those.

You're saying that one column could be being recognized as text strings,
when the other column its comparing to could be being recognized as numeric
values, and as such it returns a false because the text string is not
recognized the same as a numeric value.

Correct.

This then takes me back to the point of WHY?

Because that's how Excel works. Available data types include numbers
(usually double precision floating point), text, boolean and error
values. A value in one data type NEVER equals a value in another data
type even if they share an identical text representation.

There are some programming languages in which this isn't the case. VBA,
for instance, in which 1234# = "1234" returns TRUE. But that's because
VBA and those other languages apply implicit type conversion even to
operands of comparison operators. Excel doesn't. Whether it should or
not is a different matter, and purely academic because Microsoft is
extremely unlikely to change this behavior lest it break other people's
existing formulas.

I don't care what the value is. I don't care that 12345 is numeric in one
cell, and a text string in the other column. I just want it to be recognized
as being the same, because regardless of the format/properties behind each,
it is still just 12345.

There you're wrong. It's text in one cell and numeric in the other. You
may not appreciate the difference, and it's clear you don't want to
have to do so, but if you want to create reliable Excel formulas you're
going to have to break down and learn this distinction. To repeat from
above, this behavior is unlikely to change. Your choices are either to
learn this and accommodate how Excel works or use something other than
Excel.

I entered it as JUST 12345. I did not enter it as "12345" in one column, and
12345 in the other. Nor did I enter it as 12345, etc... ad infinitum. Which
seems to me to be strictly a background/behind-the-scenes format issue.

OK, but if one cell was formatted as Text while the other was formatted
as General, then the first will be STORED as text and the other as
numeric.

At this point I have to say I suspect you imported the list of permit
numbers from some other system. If so, it's likely that other system
stored the permit numbers as text, and Excel is respecting that other
system's data types.

So, going back to my original point--
1- is there anyway to consistently get it to work one way, without having to
spend hours looking at why it doesn't?
....

Only by always coercing both sides of your = or comparisons to
numeric or text.

3- if no one here knows the answers that I seek, who can I speak with that
will get me the correct way to handle this?

No one who knows more about Excel that I do will give you a different
answer than I have. To repeat, this is just how Excel works, and it's
unlikely to change.

Last but not least, I don't think you have insulted anybody.
Harlan never insults anybody. He just informs in a "matter of fact" way.G

--
Don Guillett
SalesAid Software

"Epinn" wrote in message
...
Hi Harlan,



I want to thank you for being detailed and taking the time to explain
SUMPRODUCT, data types, coercion etc. I learned a lot reading your posts in
this thread. It has been educational. I'll take the following advice to
heart.



--x is guaranteed to convert text representations of numbers to their
numeric values (with possible floating point rounding/truncation error)
while leaving numeric values unchanged, and x&"" is guaranteed to convert
numeric values to their text representations while leaving text as-is.




In my opinion, the fact that you are meticulous/precise on terminology is a
reflection of your expertise. I always believe that programmers do well in
Excel because writing a formula is like writing a line of code in a program.
A sense of perfection is an important quality of a good programmer.



I also want to thank you for explaining #N/A (wildcard) to me the other day.



Last but not least, I don't think you have insulted anybody.



I am going to go over your posts again and if I have any questions, I'll
start my own thread. If I do, I hope you have time to drop by.



We are lucky to have you around sharing your expertise and *generosity*.
Please don't give up on us.



I hope you still come to this thread and can read this.



Epinn


"Harlan Grove" wrote in message
oups.com...
SteveDB1 wrote...
Ok, Harlan,
I tried the last two equations you mentioned, and the first one does not
work. It only returns a zero.
(--Map!$E$4:$E$30=$A3)

or

(Map!$E$4:$E$30=$A3&"")

Simply put, I don't believe you. If you can get the result you claim to
want using formulas like =1*Map!E4 in a different range of 27 rows by 1
column and using that range rather than Map!E4:E30 in your SUMPRODUCT
formula, then Excel would give the same result using --Map!E4:E30 in
the SUMPRODUCT formula. I've NEVER seen an exception to that, and I'm
going to be skeptical about claims otherwise until I see a live
example.

Maybe you meant that Map!E4:E30 is numeric but A3 isn't. In that case
use

(Map!$E$4:$E$30=--$A3)

or

(Map!$E$4:$E$30&""=$A3)

But to be as generic as possible, use

(-Map!$E$4:$E$30=-$A3)

or

(Map!$E$4:$E$30&""=$A3&"")

So, all of this still brings me back to the point of-- how do I,
consistently, constantly, get it to work, without having to go through and
find out why one equation works and the very next one does not work.

There's NOTHING wrong with Excel's -- double operator or &"" or
SUMPRODUCT. The problem lies entirely in probable data type mismatches
between your ranges Map!E4:E30 and A3. It's entirely and exclusively up
to YOU to ensure that you're comparing text to text or numbers to
numbers.

My whole point for coming in here to ask is that the equation is not
consistently working, and I'm spending far too much time going through
looking for small, inconspicuous reasons for its failure. I came looking to
learn more about this equation because it did once work, and it sped up my
work process by an undefinable amount of time. It's now taking 3 x's longer
to figure out why it's not working.

It's not an 'equation'. It's a formula.

If your formulas appear to behave inconsistently, far & away the most
likely answer is that there are problems with YOUR data. To repeat,
it's up to YOU to ensure that you're comparing text to text or numbers
to numbers. There are ways to do that: --x is guaranteed to convert
text representations of numbers to their numeric values (with possible
floating point rounding/truncation error) while leaving numeric values
unchanged, and x&"" is guaranteed to convert numeric values to their
text representations while leaving text as-is. You could also have
trailing nonbreaking HTML spaces, and Roger Govier has already shown
how to handle those.

You're saying that one column could be being recognized as text strings,
when the other column its comparing to could be being recognized as numeric
values, and as such it returns a false because the text string is not
recognized the same as a numeric value.

Correct.

This then takes me back to the point of WHY?

Because that's how Excel works. Available data types include numbers
(usually double precision floating point), text, boolean and error
values. A value in one data type NEVER equals a value in another data
type even if they share an identical text representation.

There are some programming languages in which this isn't the case. VBA,
for instance, in which 1234# = "1234" returns TRUE. But that's because
VBA and those other languages apply implicit type conversion even to
operands of comparison operators. Excel doesn't. Whether it should or
not is a different matter, and purely academic because Microsoft is
extremely unlikely to change this behavior lest it break other people's
existing formulas.

I don't care what the value is. I don't care that 12345 is numeric in one
cell, and a text string in the other column. I just want it to be
recognized
as being the same, because regardless of the format/properties behind each,
it is still just 12345.

There you're wrong. It's text in one cell and numeric in the other. You
may not appreciate the difference, and it's clear you don't want to
have to do so, but if you want to create reliable Excel formulas you're
going to have to break down and learn this distinction. To repeat from
above, this behavior is unlikely to change. Your choices are either to
learn this and accommodate how Excel works or use something other than
Excel.

I entered it as JUST 12345. I did not enter it as "12345" in one column,
and
12345 in the other. Nor did I enter it as 12345, etc... ad infinitum.
Which
seems to me to be strictly a background/behind-the-scenes format issue.

OK, but if one cell was formatted as Text while the other was formatted
as General, then the first will be STORED as text and the other as
numeric.

At this point I have to say I suspect you imported the list of permit
numbers from some other system. If so, it's likely that other system
stored the permit numbers as text, and Excel is respecting that other
system's data types.

So, going back to my original point--
1- is there anyway to consistently get it to work one way, without having
to
spend hours looking at why it doesn't?
....

Only by always coercing both sides of your = or comparisons to
numeric or text.

3- if no one here knows the answers that I seek, who can I speak with that
will get me the correct way to handle this?

No one who knows more about Excel that I do will give you a different
answer than I have. To repeat, this is just how Excel works, and it's
unlikely to change.

Hi all.
After intense discussion, and "debate" I'll consider this post answered for
now.
I have in fact tried all of Harlan's, and Roger's recommended solutions, and
have come up with what appears to be a viable means of solving the problem.
And as such has already been implimented in updating workbooks.
However, one of my colleagues has encountered an issue that I've not
encountered, and in fact, it has never been an issue for me.
He's found that all works except when he uses a formula in the final source
column-- i.e., the last "array" of the sumproduct, in our case:
=sumproduct ((...)*(...)*(Map!$C$3:$C$n))
Where 'n' is the end of the row range, and the formula that I'm referencing
in the C column is of the general form- (=B2*x/y)-- ignore the paren's; x,
and y are some predetermined values-- you can pick any arbitrary numbers.
I quite frankly can't think of any reason why it would not work, as mine has
always worked for this portion, and never once been an issue. My issues, it
turns out were cross-datatypes that were returned as false, when the values
represented should've returned true.
So, again-- thank you for everyone's assistance.
Hope all have a great T-day-- for those on this side of the "pond". For
those overseas... enjoy the rest of your week.
See you all next time I need help.

"SteveDB1" wrote:

Hi folks.
I use sumproduct to sum columns of values after testing against 2 criteria.
At first it was working great, and then it stopped working once one cell
that I was testing against had a value in it. It's actually become quite
finicky. It'd work in some cases, but not in others.
My version of sumproduct is:
=sumproduct((Col&RowRange = cell a)*(Col&RowRange = cell
b)*(Col&RowRangeFinal))
Where Col&RowRange would be a column, and range of rows; cell a, and cell b
would be the values that I'm looking for; and the Col&RowRangeFinal would be
the column that I'm summing up.

I have to ask-- would this be affected by cell formatting? (I've gone so far
as to copy and paste cell values from the source regions to the final
regions, and it still only works sometimes.) If so, why? It'd be really nice
to know, so I can ensure consistency, and continuity.
what else would cause this to fail?
Oh, and I did try the comma's instead of *'s as shown in the office
assistant help file. That killed everything with no sign of errors, or #ref,
value, etc....-- i.e., I just got 0.000 for the returned value.

Thus far, aside from this issue, it's been a real godsend, so I'd hate to be
forced to use something else to solve my reasons for using it to begin with.

Hi Steve

Glad to hear that the explanations and formulae I sent you back having
seen your workbook solved your problems.

You don't say what problems you are now getting. What is the error that
is being returned?
Are you talking about your Acre-feet column like column D in the
workbook you sent me, with formulae like =C2*200/50 ?
On your workbook, that was accurately being calculated in column J of
sheet Sum.

If it is not working in another workbook, then there must be something
wrong with one or more of the values in column B, giving rise to an
error in the calculated result for column C.
It will not be an error with Sumproduct, but an error with the source
data that needs correcting.

--
Regards

Roger Govier


"SteveDB1" wrote in message
...

Hi all.
After intense discussion, and "debate" I'll consider this post
answered for
now.
I have in fact tried all of Harlan's, and Roger's recommended
solutions, and
have come up with what appears to be a viable means of solving the
problem.
And as such has already been implimented in updating workbooks.
However, one of my colleagues has encountered an issue that I've not
encountered, and in fact, it has never been an issue for me.
He's found that all works except when he uses a formula in the final
source
column-- i.e., the last "array" of the sumproduct, in our case:
=sumproduct ((...)*(...)*(Map!$C$3:$C$n))
Where 'n' is the end of the row range, and the formula that I'm
referencing
in the C column is of the general form- (=B2*x/y)-- ignore the
paren's; x,
and y are some predetermined values-- you can pick any arbitrary
numbers.
I quite frankly can't think of any reason why it would not work, as
mine has
always worked for this portion, and never once been an issue. My
issues, it
turns out were cross-datatypes that were returned as false, when the
values
represented should've returned true.
So, again-- thank you for everyone's assistance.
Hope all have a great T-day-- for those on this side of the "pond".
For
those overseas... enjoy the rest of your week.
See you all next time I need help.

"SteveDB1" wrote:

Hi folks.
I use sumproduct to sum columns of values after testing against 2
criteria.
At first it was working great, and then it stopped working once one
cell
that I was testing against had a value in it. It's actually become
quite
finicky. It'd work in some cases, but not in others.
My version of sumproduct is:
=sumproduct((Col&RowRange = cell a)*(Col&RowRange = cell
b)*(Col&RowRangeFinal))
Where Col&RowRange would be a column, and range of rows; cell a, and
cell b
would be the values that I'm looking for; and the Col&RowRangeFinal
would be
the column that I'm summing up.

I have to ask-- would this be affected by cell formatting? (I've gone
so far
as to copy and paste cell values from the source regions to the final
regions, and it still only works sometimes.) If so, why? It'd be
really nice
to know, so I can ensure consistency, and continuity.
what else would cause this to fail?
Oh, and I did try the comma's instead of *'s as shown in the office
assistant help file. That killed everything with no sign of errors,
or #ref,
value, etc....-- i.e., I just got 0.000 for the returned value.

Thus far, aside from this issue, it's been a real godsend, so I'd
hate to be
forced to use something else to solve my reasons for using it to
begin with.

"SteveDB1" wrote:

Hi folks.
I use sumproduct to sum columns of values after testing against 2 criteria.
At first it was working great, and then it stopped working once one cell
that I was testing against had a value in it. It's actually become quite
finicky. It'd work in some cases, but not in others.
My version of sumproduct is:
=sumproduct((Col&RowRange = cell a)*(Col&RowRange = cell
b)*(Col&RowRangeFinal))
Where Col&RowRange would be a column, and range of rows; cell a, and cell b
would be the values that I'm looking for; and the Col&RowRangeFinal would be
the column that I'm summing up.

I have to ask-- would this be affected by cell formatting? (I've gone so far
as to copy and paste cell values from the source regions to the final
regions, and it still only works sometimes.) If so, why? It'd be really nice
to know, so I can ensure consistency, and continuity.
what else would cause this to fail?
Oh, and I did try the comma's instead of *'s as shown in the office
assistant help file. That killed everything with no sign of errors, or #ref,
value, etc....-- i.e., I just got 0.000 for the returned value.

Thus far, aside from this issue, it's been a real godsend, so I'd hate to be
forced to use something else to solve my reasons for using it to begin with.

For some reason, I'm not able to see the OP, but am reading this through
Robin's post.

As in all cases, no information is passed to the group when you just say
"not working"!

You described the return of the unary version as a zero total, but you
didn't describe your problem with the asterisk version, except that it "only
works sometimes".

What happens when it (asterisk version) doesn't work?

Your zero total using the unary version leads me to believe that the data in
your totaling range are *all* text values.

This is partially substantiated by the fact that the asterisk version works
"sometimes".

The asterisk version will total numeric and text values, as long as they
even look like numbers.
The unary version will bypass anything that's text, and simply not calculate
them, with no warning or error messages.

The asterisk version, even though it will calculate almost anything that
even looks like a number, will error out if there is any value in the
totaling column that doesn't look like a number.
This includes nulls ( "" , zero length strings) that may be the returns of
formulas in the totaling column,
or invisible characters that may have been imported with the numbers from a
web site (notably Char(160)).

SO, if you're "not working", in relation to the asterisk form means a
#Value! error, look in your totaling column for non-numeric text, or web
imported characters.
--

HTH,

RD
================================================== ===
Please keep all correspondence within the Group, so all may benefit!
================================================== ===



"Robin McLean" Robin wrote in message
...


"SteveDB1" wrote:

Hi folks.
I use sumproduct to sum columns of values after testing against 2
criteria.
At first it was working great, and then it stopped working once one cell
that I was testing against had a value in it. It's actually become quite
finicky. It'd work in some cases, but not in others.
My version of sumproduct is:
=sumproduct((Col&RowRange = cell a)*(Col&RowRange = cell
b)*(Col&RowRangeFinal))
Where Col&RowRange would be a column, and range of rows; cell a, and cell
b
would be the values that I'm looking for; and the Col&RowRangeFinal would
be
the column that I'm summing up.

I have to ask-- would this be affected by cell formatting? (I've gone so
far
as to copy and paste cell values from the source regions to the final
regions, and it still only works sometimes.) If so, why? It'd be really
nice
to know, so I can ensure consistency, and continuity.
what else would cause this to fail?
Oh, and I did try the comma's instead of *'s as shown in the office
assistant help file. That killed everything with no sign of errors, or
#ref,
value, etc....-- i.e., I just got 0.000 for the returned value.

Thus far, aside from this issue, it's been a real godsend, so I'd hate to
be
forced to use something else to solve my reasons for using it to begin
with.