Print Initial Gender with airSlate SignNow

Print initial gender

hey friends in this video I'll show you how to guess the gender of a customer or user based on their first name my name is Tom I'm a data scientist in Australia and let's dive straight in so we're going to use some real data from the United States Social Security and you want to download the national data using the link in the description below so once you've downloaded the national data and you've extracted that because it's a zip file then move over to our studio and you can follow along with me so let's load up and create a new our script and we're going to load the tidy Bruce package to let us work with data ok so the next step is we're going to load in in fact let's first of all let's check out the contents of these zip file that we've downloaded so we're going to do that using the file system and then directory list looks and we'll check to see what's in that so ok we can see that within that zip file there's a bunch of files year of birth 1881 1883 etc and a rape meat okay so these are the first names and counts for people born in these years ok let's have a look at one of these files and see what it looks like so let's pick the year 2000 so let's say first names and we'll specify first than 2000 and we use read CSV to load the file so read CSV and let's specify the year 2000 so let's do that well we just happen there okay I see what's happened so if we look at if we look at the at the table we can see there doesn't have any header so we need to specify the header so it looks like it's the first name gender and the count so that will specify that using name's Cole underscore names is the first name the gender and the cat or best club birds in all right okay let's see what that did okay we're also going to specify the context is character otherwise it will try and convert to the type it guesses that it guess is that if means false and true but that's not right because gender will be if or in female male so let's load everything in and the character okay that worked so if we look at our data set we can see the first name we can see the gender and number of births that year and 2000 okay so the first thing we want to do let's just get a sense of with the breakdown of names as by gender so count gender okay so we can see that there's more female names or unique female names in the year 2000 and male names that's that's curious in itself let's have a look and see what the breakdown was so how many male births versus female Birds in that year one let's try that and then we're going to group by gender and we're going to summarize birds and egos the son of bears again and that will get the count of birds by each gender oh yes okay so we have to convert in fact up here so you notice that birds in wasn't larger than as a character as well we need to convert that to an integer so let's do that up here so we'll mutate this is an integer okay all right let's try that all right so we can see that there were slightly more male birds in 2000 than female birds interesting okay we could simplify this actually by using count and then we can do white equals birds in and that would just gender and that will do the same thing as what we did above all right so nobody does the basic counts less basic counts okay so now we want to move on to the next step which is we want to get a breakdown for each each first name how many people were born with that name who were female and how many were born with that name who were male so let's do that we can do that by in fact hmm okay let's pivot pivot all right let's do something first so we want to change gender let's count gender say female male okay we're going to change gender to be more specific so we'll call it male and female so we'll call gender we'll use case when to do this so when gender is M then change that tomorrow and then when gender equals if then use female and that should work let's try that all right fantastic so the next step the next step is we want to give it in fact let's just save that the next step is we're going to pivot so that we have first name female a male so we're going to do that by doing pivot wider pivot wider makes it data why don't have it longer makes it longer so pivot wider meant from echoes gender values from legal births em and you can see that we have their first name we have a number of people who were born in 2000 who were female with their name and number who Mel was their name okay so again I think we'll just say I think we just saved to that variable all right so feel free range chickens right so the next step as you want to calculate the proportion of people who have that first name who are female so let's do that like this so proportion and we need to calculate the proportion of by you by calculating figure I mean out of the number of people who are serial has the number who were male now let's see how that works okay now I want to check for something because I want to check to make sure that in cases where there's no male or no female people that that doesn't make that doesn't convert to an n/a or now so let's just check that okay so we can see for some of these names like Brittany Catherine and so on in these cases there were no people who were male who had that name and so the proportion female should be one but it's no because there's nothing to compare it to so what we'll do is up here we'll change will mutate finding out because your place in a is zero so that means if female is note then we'll change that to zero and if male is well we'll replace male sorry as well okay and let's run this again now everyone has a number okay so now we say that 99% of people were the first name Emily a female however let's let's filter for people who say proportion female is less than 10% all right so we see the people with the name Logan or Hunter were Tyler or Devon or Christian tend to be most often male alright so again we'll save that proportion to this variable we're building yourself as we go now we want to see what the distribution is and we can do that using G on density so look at density distribution proportion V now okay and that will let us know where our cutoff should be to get some reliable guesses as to the gender of someone based on their first name and of course I should clarify this is never going to be perfect a towards at but being directionally accurate as often enough when we're working with data so the first step here let's do a plot proportion we're gonna do a density plot okay awesome okay so we can see that there are very few people who fit into this bucket between about 0.1 or so and 0.9 who would be marked as kind of ambiguous between female or male and tested this distribution so that probably means that around the 90 percent mark we can pretty confidently say that someone is either female or male if 90% of people have that proportion so let's have a look and see what that does let's say that we create a new variable and we guess based on so we're going to guess we're going to and kind of guessed gender equals okay so when the proportion of people are female is greater than 90% we'll say that that person is probably female and when the proportion of people with their name is female is less than 10% will say that problem email okay let's see how that turns out so first of all let's count guess gender now okay so there are about 1,400 names where it tends to be unclear if any if anywhere doesn't fit in we just say unknown okay so now let's see what fits into each of those buckets that's okay okay so these people with these names tend to be most most commonly female let's have a look at mo okay and unknown so and one would be people whose names typically go either way all right so Alexis that's interesting Taylor Jordan right so these are names which may may skew more towards female or male a bit but but they're not ad clear Jamie Peyton okay all right all right now let's create a list of people and let's just see how well this performed so let's get a list of 20 random names that's Paul 20 random names so we can I'm just gonna copy and paste these ones in so we'll get these names now that's quick quotes around this okay all right so now we have a table with these first names so test names and we'll call it just first name alright so now let's try joining on the first names to test names and see you but we guess so we do it left join on first names mm by because the first name okay alright so we see there for later we infer that she's female and so on Polly Nora and so on rodolfo male LeAnn and so Mickey Mickey wasn't actually in the social security data so we don't know Billy female and so on and as a couple Valentin and Concepcion I think and Ricky where we don't know but generally speaking this approach tends to give us a pretty good guess as to the gender of someone based on their first night now what a clarifier me you would be very upfront like this is absolutely not a perfect approach but this can be really helpful if you want to guess the gender of your customers or users all right so if we're gonna clean this up a little bit and we're almost done the next step would just be to look at our first names we can just select first name yes gender and that would be and we can still use this going forward the guest gender table whoops okay so that was a quick look at how to infer guests names using our if you'd like to see further videos like this please let me know in the comments below and consider subscribing for more content like this