Database Reference
In-Depth Information
Customer
Demographics
CustomerKey
BusinessType
OwnershipType
TotalEmployees
PermanentEmployees
AnnualRevenue
AnnualProfit
YearEstablished
StoreSurface
ParkingSurface
DateFirstOrder
DateLastOrder
Customer
NewCustomers
CustomerKey
CustomerID
CompanyName
Address
PostalCode
CityKey
NewCustomerKey
CompanyName
StreetAddress
City
State
PostalCode
Country
BusinessType
OwnershipType
TotalEmployees
PermanentEmployees
AnnualRevenue
AnnualProfit
YearEstablished
StoreSurface
ParkingSurface
Fig. 9.1
Tables
CustomerDemographics
and
NewCustomers
added to the Northwind
data warehouse given in Fig.
5.4
Attributes
AnnualRevenue
and
AnnualProfit
have the following values:
1: Under $10,000
2: $10,000-50,000
3: $50,000-100,000
4: $100,000-500,000
5: $500,000-1,000,000
6: Over $1,000,000
Attribute
YearEstablished
has a range of 1950-1997. Attributes
StoreSurface
and
ParkingSurface
are expressed in square meters. Finally, attributes
Date-
FirstOrder
and
DateLastOrder
are derived from the
Sales
fact table.
9.1.2 Supervised Classification
Supervised classification
is the process that allocates a set of objects in
a database to different predefined classes according to a model built on the
attributes of these objects. For this, a database
DB
is split into a
training
set
E
and a
test set
T
. The tuples of
DB
and
T
havethesameformat,
while the tuples in
E
have an additional field, which is the
class identity
,
which stores the class of each tuple in the training set. These classes are used
to generate a model to be used for classifying new data. Once the model
is built using the training set (with labeled records), the correctness of the
classification is evaluated using the test set (with unlabeled records). Typical
uses of classification are credit approval (risk classification), marketing, health
planning, and so on.
Search WWH ::
Custom Search