As an assistant manager of an insurance company my task is the prediction of which customers are potentially interested in a caravan insurance policy based on both socio-geographic and personalized data. For model building, data of 4000 customers and 86 variables, including the target variable was available. Also, give an explanation why these customers would buy the caravan insurance company. Make my insights useful and action in order to report it to my boss with no prior knowledge of computational learning technology.
Method used
We are going to use simple models because it is not complexed and it seems to work better in our case. Given the data, we are going to use binary logistic regression to predict the solution because we have …show more content…
Variable(s) entered on step 1: PPERSAUT, MKOOPKLA, and ABRAND.
*If we look at the significance in this table for all variable is .000. This is very important and it means statistically significant factors in our model.
*If we look at the odd ratio Exp (B), the rates are all over 1 which is good, it gives an idea of the magnitude of the effect that each kind of these variables might have on predicting the outcome variable.
People interested in buying a caravan insurance policy and why.
We can conclude that people who have more than one car are more likely to be interested in caravan insurance. Since the caravan must be pulled by a car if they don’t have a car how they are going to tow the caravan. In addition, people with more than one car policy insurance could have two cars, one regular car for they daily basis use and a big car to pull the caravan.
People with fire insurance are most likely to be interested in caravan insurance because having fire insurance could be an indicator of having a caravan. We usually use liquid propane gas in order to cook in our caravan and for their safety they buy fire …show more content…
Reflection from the assignment
Personally, we enjoyed this assignment greatly and learned valuable practical lessons. We discovered the process how companies use sets of data to have some insights in order to help managers make better and valuable decisions. Also, this assignment gave me the opportunity to explore IBM SPSS environment, and the ability to read the outputs reports and interprets the results. It was a good learning experiment.
It is very hard for Managers with no statistical knowledge to understand reports that contain statistical model because they will feel lost and completely stressed, therefore, its highly recommended and practical that the descriptions and accompanying interpretation for any problem solution had to be comprehensible, useful and actionable for professional with no prior knowledge of computational learning technology. In our opinion, we have to include minimal statistical information and a good description in words in any