Statistical functions: count but disregard duplicates

Hi,

Long-time Excel user but fairly new to the statistical functions. Have a bit of a dilemma that I hope someone can help me with.

I have a spreadsheet that consists of several hundred (will eventually be about 10,000) companies and the types of products they manufacture. It is sorted by customer number, which is unique for each company. Most companies have more than one line in the spreadsheet (many may have up to 30 lines), because they manufacture the same type of product under more than one brand name AND/OR because they manufacture two or more different types of products.

Simplified example:

Customer Number Brand Type
1 A VACUUM CLEANERS
1 A TOASTERS
1 B VACUUM CLEANERS
1 C AIR CONDITIONERS
2 D TOASTERS
3 E AIR CONDITIONERS
4 F AIR CONDITIONERS
4 F TOASTERS
4 G TOASTERS

and so on.

What I would like to do is figure out how many of these companies, of the total number, manufacture a given product type, like toasters for example. If I do a straight COUNTIF on the type column, it will return 4 which is not correct. 3 of the 4 companies manufacture toasters, but since it's listed twice under company 4 (due to two different brand names) that will throw off the numbers. I need it to screen out this type of duplicate entry and return a value of 3.

I guess what I need is a formula that will assess all of company #1's entries, if "TOASTERS" is listed it counts as true, if there's no listing it counts as false, then it moves on to company #2's entry and so on.

I have looked all through Excel help and all over the Internet with no luck. Please help!

Thanks.

Originally Posted by voodoomusic

Hi,

Long-time Excel user but fairly new to the statistical functions. Have a bit of a dilemma that I hope someone can help me with.

I have a spreadsheet that consists of several hundred (will eventually be about 10,000) companies and the types of products they manufacture. It is sorted by customer number, which is unique for each company. Most companies have more than one line in the spreadsheet (many may have up to 30 lines), because they manufacture the same type of product under more than one brand name AND/OR because they manufacture two or more different types of products.

Simplified example:

Customer Number Brand Type
1 A VACUUM CLEANERS
1 A TOASTERS
1 B VACUUM CLEANERS
1 C AIR CONDITIONERS
2 D TOASTERS
3 E AIR CONDITIONERS
4 F AIR CONDITIONERS
4 F TOASTERS
4 G TOASTERS

and so on.

What I would like to do is figure out how many of these companies, of the total number, manufacture a given product type, like toasters for example. If I do a straight COUNTIF on the type column, it will return 4 which is not correct. 3 of the 4 companies manufacture toasters, but since it's listed twice under company 4 (due to two different brand names) that will throw off the numbers. I need it to screen out this type of duplicate entry and return a value of 3.

I guess what I need is a formula that will assess all of company #1's entries, if "TOASTERS" is listed it counts as true, if there's no listing it counts as false, then it moves on to company #2's entry and so on.

I have looked all through Excel help and all over the Internet with no luck. Please help!

Thanks.

=SUM(N(FREQUENCY(IF(C2:C10="Toasters",MATCH(A2:A10,A2:A10,0)),MATCH(A2:A10,A2:A10,0))>0))

ctrl+shift+enter, not just enter

I tend to agree with JB, Teethless Mama's suggested formula works great for small ranges but the MATCH function particularly across 10000 cells is going to be slow.

Here's another alternative using Database function DCOUNTA.

I put "Type" in E1 (this has to match the header in C1) and E2 holds the Product you want to count, e.g. TOASTERS.

F1 has a header which isn't the same as any of the headers in A1:C1, I chose "Unique" and under that in F2 this formula

=SUMPRODUCT(--(A$2:A2=A2),--(C$2:C2=C2))=1

[Note this cell will dispaly TRUE or FALSE, it doesn't matter which]

Now for the unique count in E4 use this formula

=DCOUNTA(A1:C10000,1,E1:F2)

just change E2 to whatever you want to count......see attached