Allstate Auto Insurance Purchanse Recommendation Project Report
[pic 1]Allstate Auto Insurance Purchase Recommendation Project Report  BYGB 7967 Data Mining For Business Section 2Fantastic Four Group Member:Jiacheng ChenTaineng LuHaotian XiaChenliang ZhuAbstract:        With hundreds of auto insurance companies and thousands of insurance coverage options, customers often go through a lengthy process when choosing the appropriate insurance option that best suits their needs. By adding efficiency to customers’ purchasing process, insurance companies will be able to serve more customers, which translates to increased insurance sales.Introduction:          Founded in 1931, The Allstate Corporation, headquartered in Northbrook, Illinois, is one of America’s largest publicly held personal lines property and casualty insurer, serving approximately 16 million customers across the country with various product lines, including auto, house and life insurances, etc.         The topic for this project is the “Allstate Auto Insurance Purchase Recommendation”. Customers often go through lengthy processes when comparing auto insurance coverage options from different insurance providers. This not only causes Allstate serving fewer customers, but also adds a risk factor where potential customers choose a different insurance company during this process.
The goal for this project is to categorize Allstate’s potential customers into four clusters based on their demographic and vehicle features and to recommend the pre-designed coverage packages to each cluster that is most likely to purchase. This will increase the number of potential customers served by Allstate, which ultimately increase auto insurance sales. Data Description:         The dataset was obtained from “Allstate Purchase Prediction Challenge” on Kaggle.com. The dataset contains 665,249 individual cases with Customer ID as the unique identifier for the customers in the dataset. All other attributes in this dataset can be found blow in variable descriptions table.        In this dataset, each customer has many shopping points, which is defined by a customer with certain characteristics viewing a product and the product’s associated cost at a particular time. Moreover, each product has seven customizable options which can be selected by customers based on the customers’ demographic characteristics and vehicle conditions.Variable DescriptionsCustomer_IDA unique identifier for the customerShopping_ptUnique identifier for the shopping point of a given customerRecord_type0 = shopping point, 1 = purchase pointDayDay of the week (0-6, 0 = Monday)TimeTime of the day (HH:MM)StateState where shopping point occurredLocation Location ID where shopping point occurredGroup_sizeHow many people will be covered under the policy (1,2,3 or 4)HomeownerWhether the customer owns a home or not (0 = no, 1= yes)Car_ageAge of the customer’s carCar_valueHow valuable was the customer’s car when newRisk_factorAn ordinal assessment of how risky the customer is (1, 2, 3, 4)Age_oldestAge of the oldest person in customer’s groupAge_youngestAge of the youngest person in customer’s groupMarried_coupleDoes the customer group contain a married couple (0 = no, 1= yes)C_previousWhether the customer has purchased the product option C (0 = no, 1= yes)Duration_previousHow long (in years) the customer was covered by their previous issuerA,B,C,D,E,F,GThe coverage optionsCostCost of the quoted coverage options