Sign West Virginia Affidavit of Identity Fast

Document type sign affidavit of identity west virginia fast

[Music] in this demonstration we're going to talk about how states can leverage data that's already at their disposal to fight against common types of unemployment fraud there are many ways of committing unemployment for odds and several techniques of committing unemployment fraud that we're going to talk about today so let me take you through what you're looking at this is what we call a data model and it exists inside a sandbox type environment each one of these icons represents a data set and the data is all simulated but it should be data that approximates data that each state agency has at their disposal in the center we see people receiving benefits these are the people that are receiving state unemployment benefits on some sort of regular basis these payments indicate the payment amounts frequency dates things like that that those individuals are paid the receiving accounts are the accounts that uh those deposits are made into states this is just a table of all 50 states but i put this here to to emphasize the fact that data walk can be collaborative you don't necessarily need one instance of data walk for one agency multiple states can go in and all pipe their data in together into one model this will allow you to look for different types of fraud the master death index is a list available from the social security administration to any state agency that requests it this is just first name last name date of birth and some other critical information of everybody who has had a death certificate issued for them individuals who are in prison should be available in a state inmate database and individuals who have state warrants should also be available in this databases individuals who are deceased or inmates make excellent targets for identity theft once the identity is of an individual stolen the fraudster will then file for unemployment benefits in their name and most of those frauds aren't uncovered until the individual in question gets notified that there are now benefits being paid to them and they in turn notify the social security administration or the state agency hey i never filed for this what's going on when individuals are deceased or in prison obviously they're not able to make uh to receive those in those notifications and they probably aren't going to be communicating with social security administration so they make great targets for identity theft which in turn their identities get used for unemployment fraud payroll tax payments every state is a little bit different but most states have some sort of system whereby employers have to make payments either bi-weekly monthly or quarterly for everybody that they employ one of the most common types of payroll fraud are individuals who are actually still gainfully employed filing for unemployment then they receive their income from their job and also unemployment benefits from the state we'll talk about how to leverage these data sets together to flag individuals who may be committing that type of fraud and then of course we've got home addresses these are uh home addresses of individuals receiving those benefits and we've parsed out states from that to allow us to throw some filters in later so let's get right into developing analysis with data walk leveraging data sets against each other in data walk is very simple it's just a matter of clicking so for instance if we were to look at states this tells me that there are 50 states in the state database and those states in this particular database have made 3 000 payments to 1 000 individuals one way of committing unemployment fraud is to file for unemployment in multiple states in the wake of coronavirus with the crushing torrents of unemployment claims many states have relaxed the paperwork standards for and their cross-checking standards to ensure that the sort of fraud doesn't happen but looking for individuals who are receiving payments from multiple states is easy all we need to do is click on this connection and say show me anybody who is connected to two or more states and as you can see our people receiving benefits have moved from a thousand to one we want to investigate that individual we can pull up a table and see his information or we can just go ahead and throw him on a link chart so barclay qrikmir with the date of birth of september 2nd 1967 is connected to multiple state payments and let's look at his linked objects first of all let's check out those payments see what he's getting so far he's had three it looks like they started in march of 2020 and now let's see what states are actually sending those payments it looks like he's gotten two payments so far from virginia and a third from maryland if we want to more quickly summarize this individual especially if he has lots of payments it's easy to do so we can start here and add a second degree connection to the state and we can tell it summarize all of those payments to give us a count of how much money each state has paid him we can then hide these guys we see that virginia has paid this individual 1593 dollars and marilyn has paid him almost 900 we can save this analysis if we'd like let's go and save this and say individuals connected to more than one state data walk will automatically pull new data into the system at regular intervals now that i've created this analysis and saved it no one ever has to come in and click from icon to icon or set those filters ever again this analysis is going to be saved in my workspace so now whenever new data pulls in not only will the universe viewer with all of those different data sets update but so will this analysis here we go so if i want to see the results of this tomorrow after data is updated i don't have to go into the universe viewer i can just pull up the results right here from my workspace additionally you may be in a situation where somebody comes to you with a watch list and says you know let me know if any of these individuals show up in our data or let me know the minute somebody gets a payment from more than one state i don't have to come in and rerun this analysis i don't even have to necessarily come into my workspace and pull up the analysis from this workspace i can configure an alert so i can say we'll call this individuals connected to more than one state notify me if there are any changes to the results of the analysis send me an email and then one state notify officer smith so now if data gets updated every night at 9 00 pm data walk is going to ingest that new data it's going to update all these analysis and if the results of this change it's going to send me that email and then i'm going to be able to instantly notify officer smith hey we've got a new individual who's connected to more than one state by payment additionally this analysis every step that i've taken has been saved in what we call breadcrumbs this is this light gray bar at the top left so if there's ever any question about how it is we arrived at the results that we arrived at we've got everything saved and data walk will do this during every step of every analysis that you do it will automatically record all of your steps so we've investigated an individual who's receiving payments from more than one state we could also investigate individuals who are receiving payments from outside of the state in question so let's just say that we work inside i don't know the state of virginia so all these people should be virginia residents we can connect them to their home addresses and to our home states and now let's add a filter and let's say again we're interested people who don't live in virginia let's exclude any address that's inside virginia to get us only addresses that are outside looks like we've got 354 addresses connected to people receiving benefits from inside virginia now if we wanted to we could very easily check these out on a map and we could very quickly see that each of these individuals who are receiving payments from the state of virginia are attached to addresses outside of virginia and sure enough you see a lot of addresses in maryland some in west virginia that might be indicative of fraud or it might not and again if we want to save this analysis and see who those people are i'm going to come up here i'm going to say individuals out of state and we can reset and we can look for another type of fraud let's look for individuals that are in prison and check out their payments so again we're going to go to people receiving benefits we're going to see if any of them match on first name last name and date of birth of anybody who's currently being held in the state prison and sure enough we've got five let's throw those five people over on a link chart and see what we can find out about them so the first thing we want to do is make sure that they match the individuals who are in prison and we can see pin down goody with a date of birth at 10 11 1999. sure enough penn dowgity has taken in on one two 1998 if this is looking a little messy we can rearrange it however we might like let's see where these individuals are being held you can throw the prison in see that four of them are being held at baskerville correctional center one others held being held at augusta this is an important insight too if you have a lot of individuals who are in prison who are receiving these benefits uh you know maybe they were started receiving the benefits and then were arrested and went to prison and we don't necessarily have any fraud but if you have a lot of them that are connected to one prison and none that are connected to others you might have a data breach at that prison the prison might hold data that fraudsters have now stolen to take out fraudulent insurance payments in these individuals names so that's something to look out for so we can see when they were taken in and what for let's see when these payments were made and we can clean this up a little bit down the state all right so they're all in virginia and let's redo this with virginia at the top so now i can examine all of these and i can say all right penn delgadi he was receiving a payment june uh 22nd of 2020 but it looks like he was taken into prison in 1998 for manslaughter so that's definitely something that's going to be suspicious that we're going to want to look into a little bit more so we can sort of visually go through these and make sure that none of these payments were occurring before they went to prison see which ones started coming in after they went to prison things like that again if we want to save this analysis we can we'll call this individuals in prison now i've already made a few analyses you saw me click back and forth between individuals and inmates to do the same thing for payroll tax payments warrants and death indexes all we would do is click on our people receiving benefits click over to the master index and click back and see how many matches there are so i've already created those analysis we're going to skip that part and we're going to move into a risk model now there are some things that we know about identity theft victims the elderly and the very young tend to be at very high risk so anybody who's over the age of 55 anyone who's under the age of 20 is at higher risk of identity theft we also know that payments from unemployment benefits are based on how much you were making before you went on unemployment you know there's a whole range of possibilities how much those payments might be if we want to sort of study the distribution of payments in this data set all we have to do is go over to our payments open up our table and click over to the amount column and here we can tell it to give us some summary statistics so we can see a minimum of seven hundred dollars in each payment a maximum of nine hundred dollars and an average of seven hundred and ninety dollars so this is a fairly normally distributed payment structure but as a fraudster there's no reason to be honest about your income before you file your claim so for officers are more likely to claim something near the maximum something near 900 so there are high risk payments and individuals receiving those high risk payments so let's try to find out who those individuals are so the first thing i'm going to do is put a filter on payments and i'm going to come down here to amount i'm going to say anything greater than 880 there's 219 payments of 880 or more made to 201 people so we'll call these high amount payments now we've made a number of analyses that link high risk factors back to individuals receiving payments your state might have tens of thousands of individuals if not millions of individuals who are on unemployment and you don't necessarily have the resources to investigate all of them so prioritizing them by using these analysis is great but we could also combine all of these analyses into a risk score that might help us more efficiently use our time to do so in data walk is easy so we're going to decide which people it is we want to investigate and we're just going to come down here and add a score the first thing we'll need to do is add a name and we'll call it unemployment risk this is going to calculate a score for every person receiving an unemployment payment and we have to weight that score by using the analyses that i've already uh put in so here's our high amount payments you know maybe we want to give that a i don't know a one individual's connected to more than one state that's pretty suspicious maybe we'll give that a three individuals in prison similar very suspicious maybe we'll give that a three individuals out of state there are i suppose conditions where someone could live out of state and still be receiving in-state benefits individuals over 55 that certainly puts them more at risk individuals over 20. unemployment with death certificates these are individuals who are receiving unemployment benefits yet have a first name last name and date of birth match with somebody who has a death certificate issued to them so we can give that three unemployment with payroll tax payments these are individuals who are receiving unemployment but also had an employer make a payroll tax payment for them recently and we'll give that a three and unemployment with warrants i'll give that a one the great thing about risk scores and data walk is that there is no limit to how many you can calculate there will always be some discussions inside your group about well you know i think that this variable deserves more weight than this one and we should adjust this what's great is that you can calculate as many risk scores as you want you can compare them over time you can find which ones are more accurate which ones produce fewer false positives and you can begin to sort of tune your risk score practices as you get more data so as you can see it's going to go ahead and connect a new data set to our people receiving benefits if we click on over to it it's going to tell me that 614 people received a non-zero risk score and if we open up the table view it's going to show us each individual and why they received the score they got so this individual got a score of one because he has a home residence out of state this individual got a four because they're over 55 and because they are receiving unemployment benefits while also having an employer make payroll tax payments for them so maybe you decide uh great we'll just investigate these individuals with risk scores or maybe you want to set some further criteria so we could say show us anybody with a score greater than or equal to five it's going to dramatically reduce our workload down to 19. we go i can pull up those individuals and i can investigate them one one at a time using a link chart if i want there's mike rent come through his home address on there i can throw his payments on there maybe he's connected to a prison i can show his accounts i can show what state those payments came from we can see virginia's been sending him payments since about may he got a payment on may 1st may 17th may 24th i can look and see where his home address is on a map it is over here looks like just across the state line into west virginia if i decide this individual is a target and i want to present this information i have a couple of different ways of doing it first of all can mark him mark him red let him know he's the target i can put him at the top i can export this image as a png as a jpeg as bitmap to embed it into an intelligence report i can pull up his folder which will serve as a mini intelligence report it shows all of his critical information all of the data sets that he's linked to his states his home addresses his payments his accounts i can export this as a pdf show all those links this is completely customizable if you've got photos in your database we can throw a photo in there we can create templates for individuals and it may be another one for businesses but this report can be configured however you would like here's our pdf for mike rentcom we see the payroll tax payments that have been made for him his bank accounts the states that are giving him payments his address his risk score all of that information is at our fingertips we also may want to present this to somebody and i find that the history at the very bottom is a really easy way to do that history serves two roles one it allows you as the analyst to sort of go back to an earlier version of your link chart so if you find yourself adding too much data and it becomes convoluted and you say well hold on i want to back up and start from an earlier period you can but if you present these findings you can also use this sort of as a presentation and you can say all right uh we developed a risk score mike grint was at the very top so we began looking at him i added his accounts uh his payments notice he has an address in west virginia i took another step and saw that virginia's the one paying him here's a map where he lives here's all of his information and you've already got the print out that you can hand we can also transform data to find new and emerging techniques of committing unemployment fraud so for instance many states mandate that you have to give a valid email address whenever signing up for unemployment benefits and email validation is commonly used in a lot of different industries however fraudsters are taking advantage of a little known exploit in gmail accounts gmail and google in general disregards any dot inside an email account name so john.smith gmail.com is recognized as the same thing as john smith all one word gmail.com or j.ohnsmith gmail.com they're all the same however standard data analysis practices are going to recognize those as different email addresses even though in effect they're not so knowing that it's really easy to transform data inside data walk to find things like that so if we open up our table view of people receiving benefits we can see everybody is related to an email address here they are and we scroll down just a little bit i've embedded a fraudster kcaslys gmail.com you'll notice all five of these are distinct email addresses with different dots all throughout now if we were to say throw a histogram together and look for duplicate email addresses in something like microsoft excel you would have to import data create some calculations and then build a histogram around it here it's just one button and we can look at all the emails inside this data set and we can see all right well the top four most common only occur once what that's telling me is that no email address occurs more than one time but we know about this gmail exploit so what we want to do is open up our table and transform all of this data to clean it up to look for duplicate email addresses so all we're going to do is go to our data transformations we're going to show advanced options and we're going to select replace a dot with nothing and apply data walk is going to automatically transform that now when we look at our histogram it's not going to be able to hide behind those multiple dots okay casually gmail.com is going to appear five different times so it's just that easy to develop multiple analysis to leverage different databases together to use those analyses to create risk scores to investigate individuals once you've found high-risk transactions and to even transform your data with just a few clicks to address new and emerging trends inside fraud [Music] you