Insurance agent expectations and productivity

 in there over 300,000 independent agencies in the United States not just the 21 states that Westfield operates in but Nationwide so how can we take to the third-party David and bridges the Gap that we currently have to go for what we know as certainty about the agencies today to what we what we can expect from those prospects in the future world see using collaborative filtering then to give the decision-maker some other measure a page this agency is very similar to these other five agencies you currently have the relationships with and exposure to and you're in your Fleet today and here cells similar that they are to give them a better understanding then of how well that agency will fit with the current Fleet if they were to be appointed and then all of that all the data prep from our prepare process as well as the Dayton 20 ounces from the prescribed in the Predictive Analytics outputs from the collaborative filtering logistic regression feed the previous present process to enable fact-based decision-making so in more detail now and each of those processes so we have we had all they had all this data but what can we do with it we had third-party data to have been purchased on attributes of agencies as well as agent prospects within the footprint we had information on her agencies where are they located what policies have they sold where those policies located within those geographies how many people rent houses vs. owning how many Autos are there that could potentially be insured and how much the income does that Mark and have such that an individual be able to purchase a home or an auto policy and so we're going to take all that dated then he cleans it and and prepare it based on various case whether that's the agency's code or V code of the GI of the county which they're located or that policy is located in we're going to determine the market for each area but then it went well doing that were simultaneously return to the table to do some geospatial analytics to determine what their primary County what is their market share within that primary County with in Westfield agencies for all of the counties in their Market how much of that market share do they have and then plot that as well giving it Westfield 2 for the first time a visual on each individual agency's marketsandmarkets size I'll put your not then we're going to have a What are going to be our inputs as well as our outputs for our date envelopment analysis and then we're also going to have the independent variables for our product process as well the prescribed process providing the dependent variable that will mean which is that if fish and see score to use as the dependent variable in our Logistics regression down are prescribed process so we have all these different measures we have what the premium is new business what's renewal business what home is Home Premium is new business versus renewal business how do we evaluate multiple metrics simultaneously also taking into account what they had at their disposal to create those outputs to our performance are those 4 different versions of written premium and the resources they have at their disposal than her how many all those how many homeowners and renters we had the household income as well as how many agents at that agency are appointed to sell a license to sell then Westfield products their mathematical approaches used 8 envelopment analysis in order to be able to do that and what we identify them as our ad agencies that end up using an output orientation at 4 add a development announces what agencies in the pond are efficient Frontier building gets scraped off and hand it off for predict process those are those of the model agencies that