Determining Gender from Name 2

[SGLong]

Is anyone aware of a data file, tool or algorithm that I could use to determine a gender from a name? For example, if I were to pass this object "Steven" it would know "Male", and if I were to pass it "Stephanie" it would know "Female". There are a lot of names that are "gender neutral" like Chris, Jan, Alex, etc. I would hope that this object would return a blank, or zero, or "Unknown".

I know it won't be perfect, but if it can determine 80% of the genders correctly that's 80% that someone does not have to review.

Steve

 

[TamarGranor]

I'm afraid this is probably a fruitless pursuit. It's amazing how many names can belong to either gender, even names that we think of as clearly one or the other. For example, the actress who played the mother on The Waltons was Miss Michael Learned. I know at least one male Beverly.

All that said, you can find lists of the mens' and womens' names from the 1990 US census at

http://www.census.gov/genealogy/names/names_files.html.

It's easy to grab the text and throw it in tables. From there, you can pull out the ones that appear in only one table or the other and go from there.

Tamar

 

[Mike Lewis]

Steve,

Yes. I have used a product called ActiveGender, from The Software Company (

http://www.softwarecompany.com).

It is an ActiveX control, and works well from within VFP. Obviously, it can't be 100 percent accurate, but it does produce pretty reliable results most of the time.

Mike




__________________________________
Mike Lewis (Edinburgh, Scotland)

My Visual FoxPro site: www.ml-consult.co.uk

 

[SGLong]

Mike (Lewis) -- The ActiveGender product looks very comprehensive, but at $349 it's a little over my budget for just genderizing names. Maybe if I needed the other functionality of it the price wouldn't seem so bad.

I think between the census lists that Tamar pointed me to along with Mike Yearwood's code I have a pretty good alternative.

Thanks all,

Steve

 

[Dan Freeman]

ActiveGender says it uses a dictionary of 100,000 names. That's simply not enough.

In a previous job, I worked for a company that collected/combined publicly available data. We had somewhere around 700 TERABYTES of data online at any given time.

The best we could do was "likely gender" (that was the field name) based on spending habits and other demographic indicators. We could never get closer than a 99.5% success rate. With less data going into the determination, the accuracy goes down dramatically.

However, for some needs close enough is good enough.

 

[dbMark]

Going forward, see if for future additions can ask the applicant or submitter for their gender. You may not be able to do that for the existing database, but surely it can be added to the form.

 

[alvechurchdata]

We could never get closer than a 99.5% success rate.

That's amazingly good - how on earth did you manage it?

Surely there are more than 1 in 200 who have genuinely ambiguous names like "Chris", "Pat" and "Robin" and the situation must get worse if you're having to deal with funny spellings and diminutives like "Andy", "Bobby", "Carol", "Donny", "Eddy" etc.

I believe that German law insists that you register a baby with a name which clearly identifies its gender.

Geoff Franklin

http://www.alvechurchdata.co.uk

 

[Mike Lewis]

Dan,

In my application, the success rate was nearer 90 percent -- and we thought that was pretty good. I can understand Steve rejecting the product because of its price, but for my client it was a good investment.

Geoff,

And did you know that in the USA, Tracey is sometimes a boy's name? (Not joking.)

Mike

__________________________________
Mike Lewis (Edinburgh, Scotland)

My Visual FoxPro site: www.ml-consult.co.uk

 

[TheRambler]

>And did you know that in the USA, Tracey is sometimes a boy's name? (Not joking.)

You can find a lot about the names of babies born in the United States using the Name Voyager, you will see that Tracey was most popular as a boy's name in the sixties (and so was Lewis, long before that).

 

[Mike Lewis]

TheRambler,

That's a very cool web site. Thanks for the link.

Good to see I was so far ahead of the crowd.

Mike


__________________________________
Mike Lewis (Edinburgh, Scotland)

My Visual FoxPro site: www.ml-consult.co.uk