the model's prediction for this customer. As such, it is useful to rescore
customers in real-time based on current data. This can be achieved
using the JDM single record apply capability, designed to provide
real-time response. Refer to [Middlemiss 2001] for an actual banking
industry use case involving real-time scoring.
Problem Definition: How to Reduce Processing Time
of Residential Real-Estate Appraisals?
ABCBank wants to supplement the conventional real-estate
appraisal process by automated property evaluation to the current
system. Its objective is to reduce the processing time of home
mortgage-based loans and improve the customer experience. Refer to
[Rossini 2000] for a thorough study on real-estate value prediction.
Solution Approach: Property Value Prediction
ABCBank has accumulated data for the past year on real-estate
appraisal values for loans they process. In addition to appraisal values,
ABCBank acquired the real-estate details such as year built, home
size, features, and location details . Using the data mining regression
function, the bank can predict the property value, based on recent
trends and sales data. Depending on the confidence associated with
the prediction, where confidence refers to the width of the interval
around the prediction, the bank may heavily rely on the predicted
value in making a loan decision, which can reduce appraisal
evaluation time significantly. As real-estate value is a continuous
number, the regression function is the right technique to use. Refer to
Section 4.3 for an introduction of the regression function.
Data Specification: REAL_ESTATE_APPRAISALS Dataset
Concepts of physical and logical data specifications discussed in
Section 7.1.3 are the same for all data mining techniques. So in this
section we will not repeat the concepts covered earlier; instead we
illustrate the selected attributes for this problem and their logical
characteristics in Table 7-7. This table includes predictor attributes