ORACLE Data Mining

1

ORACLE Data Mining Included with ORACLE 10g Freely available for UVic and for

personal unlimited evaluation use.

Choose the enterprise version!

2

Creating a data mining user A data mining user is a canonical user, but with a few

more privileges than usual. As ‘system’ create a canonical user:

CREATE USER dmuser1 IDENTIFIED BY dmpswDEFAULT TABLESPACE usersTEMPORARY TABLESPACE tempQUOTA UNLIMITED on users;

Then, grant privileges to this user: GRANT create procedure to DMUSER1; GRANT create session to DMUSER1; GRANT create table to DMUSER1; GRANT create sequence to DMUSER1; GRANT create view to DMUSER1; GRANT create job to DMUSER1; GRANT create type to DMUSER1; GRANT create synonym to DMUSER1; GRANT execute on ctxsys.ctx_ddl to DMUSER1;

Now, DMUSER1 is ready to perform data mining tasks.

3

‘sh’ user To use the examples

you have to activate the sh user. There are the

example data

Created during ORACLE installation.

However, it’s not active.

To activate: Open Oracle Enterprise Manager. Login as ‘system’. Go to ‘administration’ tab, then click on users, find the ‘sh’ user and activate it.

4

Using the data of ‘sh’ Connect (login) as sh and grant SELECT privilege to

dmuser1 for these tables: COUNTRIES CUSTOMERS PRODUCTS SUPPLEMENTARY_DEMOGRAPHICS SALES

In other words execute: GRANT SELECT ON customers TO dmuser1; GRANT SELECT ON sales TO dmuser1; GRANT SELECT ON products TO dmuser1; GRANT SELECT ON supplementary_demographics TO

dmuser1; GRANT SELECT ON countries TO dmuser1;

Now, dmuser1 is able to execute SELECT on these tables, and create views on them.

5

DMUSER1: Creating views on ‘sh’ data

CREATE VIEW mining_data_build_v ASSELECT a.CUST_ID, a.CUST_GENDER, 2003-a.CUST_YEAR_OF_BIRTH AGE, a.CUST_MARITAL_STATUS, c.COUNTRY_NAME,

a.CUST_INCOME_LEVEL, b.EDUCATION, b.OCCUPATION, b.HOUSEHOLD_SIZE, b.YRS_RESIDENCE, b.AFFINITY_CARD, b.BULK_PACK_DISKETTES, b.FLAT_PANEL_MONITOR, b.HOME_THEATER_PACKAGE, b.BOOKKEEPING_APPLICATION, b.PRINTER_SUPPLIES, b.Y_BOX_GAMES, b.OS_DOC_SET_KANJIFROM sh.customers a, sh.supplementary_demographics b,

sh.countries cWHERE a.CUST_ID = b.CUST_ID AND a.country_id = c.country_id AND a.cust_id between 101501 and 103000;

You can find other view creation statements in dmsh.sql that will be in the web.

6

Oracle Data Miner GUI It’s a Java application. Connects through JDBC to the ORACLE database. Installed in SENG Windows lab. Can also be freely downloaded from Oracle.

7

Connecting

8

An electronics store chain wants to distribute a discount card to its customers, but only to those customers who are expected to increase their buying (and thus the company’s revenue) because of this card.

A test campaign was run on a sample of customers and the results were compiled into a table containing the customer demographics, purchasing patterns, and a measure of revenue produced by each customer.

The Problem

9

Structure

10

A fragment of data

11

Getting a feeling about the data

12

Build a classifier I

13

Build a classifier II

14

Build a classifier III

15

Build a classifier IV

I am selecting

this

16

Build a classifier V

17

Build a classifier VI

18

Build a classifier VII

19

Build a classifier VIII

P(A=1|…) will be used for sorting

tuples, as opposed to P(A=0|…).

20

Build a classifier IX

21

Build a classifier X