The Data Mining Process

Download Report

Transcript The Data Mining Process

Chapter 2
1
Business Problems and Data
Science Solution
2
Fundamental concepts
 An important Principle of data science is that data
mining is a process with fairly well-understood stages.
 Some involve the application technology, such as the
automated discovery and evaluation of patterns
from data, while others mostly require an analyst’s
creativity, business knowledge, and common sense.
3
Fundamental concepts
 Since the data mining process breaks up the overall
task of finding patterns from data into a set of welldefined subtasks, it is also useful for structuring
discussions about data science.
 This chapter introduces the data mining process, but
first we provide additional context by discussing
common types of data mining task.
4
From Business Problems to Data Mining
Tasks
 Each data-driven business decision-making problem
is unique, comprising its own combination of goals,
desires, constraints, and even personalities.
 The solutions to the subtasks can then be composed
to solve the overall problem. Some of subtasks are
unique to the particular business problem, but others
are common data mining tasks.
5
From Business Problems to Data Mining
Tasks
 Example: telecommunications churn problem(電信客
戶流失) is unique to MegaTelCo:
 Estimate from historical data the probability of a customer
terminating her contract shortly after it has expired.
 Despite the large number of specific data mining
algorithms developed over the years, there are only a
handful of fundamentally different types of tasks
these algorithms address.
 Individual(個體):entity. Ex: a customer or a business.
 Correlations(相關性):between a particular variable
describing an individual the company after their contracts
expired.
6
Classification
 Classification(分類) and class probability estimation
attempt to predict, for each individual in a
population, which of a (small) set of classes this
individual belongs to.
 Example question: “Among all customers of
MegaTelCo, which are likely to respond to a given
offer?”
 Two classes: will respond and will not respond.
 Scoring or class probability estimation
 Score representing the Probability( quantification of
likelihood)
7
Regression
 Regression(回歸)(“value estimation”) attempts to
estimate or predict, for each individual, the numerical
value of some variable for that individual.
 Example question: “ How much will a given customer
use the service?”
 Classification predicts whether something will happen.
 Regression predicts how much something will happen.
8
Similarity matching
 Similarity matching(相似度配對) attempts to identify
similar individuals based on data known about them.
 Ex: IBM is interested in finding companies similar to
their best business customer, in order to focus their
sales force on the best opportunities.
 Recommendations
9
Clustering
 Clustering(群集) attempts to group individuals in a
population together by their similarity, but not driven
by any specific purpose.
 Example question: “ Do our customers form natural
groups or segments?”
 Decision-making processes.
10
Co-occurrence grouping
 Co-occurrence grouping(共生分群) attempts to find
association between entities based on transactions
involving them.
 Example question: “ What items are commonly
purchased together?”
 Ex: analyzing purchase records from a supermarket.
 Recommendation system
11
Profiling
 Profiling(剖析)(also as behavior description(型為描述))
attempts to characterize the typical behavior o an
individual, group, or population.
 Example question: “ What is the typical cell phone
usage of this customer segment?”
 Profiling is often used establish behavioral norms for
anomaly detection applications.
 Fraud detection and monitoring for intrusions to computer
systems.
12
Link prediction
 Link prediction(連結預測) attempts to predict
connections between data items, usually by
suggesting that a link should exist, and possibly also
estimating the srength of the link.
 “Since you and Karen share 10 friends, maybe you’d
like to be Karen’s friend?”
13
Data reduction
 Data reduction(資料縮減) attempts to take a large set
of data and replace it with a smaller set of data that
contains much of the important information in the
larger set.
 Ex: a massive dataset on consumer movie-viewing
preferences may be reduce to a much smaller
dataset revealing the customer taste preferences.
14
Causal modeling
 Causal modeling attempts to help us understand
what events or actions actually influence others.
 Ex: consider that we use predictive modeling to
target advertisements to consumers.
15
Supervised vs. Unsupervised Methods
 Unsupervised: no specific purpose or target.
 “Do our customers nataturally fall into different group?”
 Supervised: specific target defined.
 “ Can we find groups of customers who have particularly
high likelihoods of canceling their service soon after their
contracts expires?”
16
Supervised vs. Unsupervised Methods
 Supervised data mining: there must be data on the
target.
 Tow subclasses of supervised data mining:
 Classification
 “ Will this customer purchase service S1 if given incentive I?”
 “Which service package( S1, S2 , or none) will a customer likely
purchase if given incentive I?”
 Regression
 “How much will this customer use the service”
17
Supervised vs. Unsupervised Methods
 A vital part in the early stages of the data mining
process
 To decide whether the line of attack will be supervised or
unsupervised.
 If supervised, to produce a precise definition of a target
variable. This variable must be specific quantity that will be
the focus of the data mining.
18
Data Mining and Its Result
 Distinction pertaining to mining data:
 Mining the data to find patterns and build models.
 Using the results of data mining.
 Churn example.
19
Data Mining and Its Result
20
The Data Mining Process
 Cross Industry
Standard
Process for
Data Mining
process.
21
Business Understanding
 It is vital to understand the problem to solved.
 A part of the craft where the analysts’ creativity plays
a large role.
 The design team should think carefully about the use
scenario.
22
Data Understanding
 The data comprise the available raw material from
which the solution will be built.
 Estimating the costs and benefits of each data
source and deciding whether further investment is
merited.
 Ex:
 Credit card fraud
 Medicare fraud
23
Data Preparation
24
Data preparation
 Often proceeds along with data understanding.
 Ex.
1. converting data to tabular format.
2. removing or inferring missing values.
3. converting data to different types.
25
Data preparation
 Leaks
a variable collected in historical data gives information on
the target variable-information that appears in historical
data but is not actually available when the decision has to
be made.
Leakage must be considered carefully during data preparation.
26
Modeling
 Output of modeling is some sort of model or pattern
capturing regularities in the data.
27
Evaluation
 Assess the data mining results rigorously and to gain
confidence that they are valid and reliable before
moving on.
 Includes both quantitative and qualitative
assessments.
28
Deployment
 Put into real use in order to realize some return on
investment.
 The clearest cases of deployment involve
implementing a predictive model in some information
system or business process.
 ex. Churn example
29
Deployment
 The data mining techniques themselves are
deployment.
 Two reasons
1. the world may change faster than the data science team
can adapt, as with fraud and intrusion detection.
2. a business has too many modeling tasks for their data
science team to manually curated each model individually.
30
Deployment
 Can also be mush less “technical”
 It is not necessary to fail in deployment to start the
cycle again. The Evaluation stage may reveal that
results are not good enough to deploy.
31
Implications for Managing the Data
Science Team
 It is tempting - but usually a mistake - to view the data
mining process as a software development cycle.
 Software skills versus analytics skills
32
Other Analytics Techniques and
Technologies
 Present six groups of related analytic techniques.
 Comparisons and contrasts with data mining.
 Data mining => automated search for knowledge,
patterns, or regularities from data.
 Business analyst => to recognize what sort of analytic
technique is appropriate for addressing a particular
problem.
33
Statistics
 Two different uses in business analytics.
1.
used as a catchall term for the computation of particular
numeric values of interest from data.
2.
denote the field of study that goes by that name.
34
Data Querying
 A specific request for a subset of data or for statistics
about data, formulated in a technical language and
posed to a database system.
 Differs fundamentally from data mining in that there is
no discovery of patterns or models.
 Ex: select * from customers where age >45
= ‘m’ and domicile = ‘ne’
and sex
35
Data Querying
 On-line Analytical Processing (OLAP)
easy-to-use GUI to query large data collections
 Data mining tools generally can incorporate new
dimensions of analysis easily as part of the exploration.
36
Data Warehousing
 Collect and coalesce data from across an enterprise,
often from multiple transaction-processing systems,
each with its own database.
37
Regression Analysis
 This will involve estimating or predicting values for
cases that are not in the analyze data set.
38
Machine Learning and Data Mining
 A field of study arose as a subfield of Artificial
Intelligence, which was concerned with methods for
improving the knowledge or performance of an
intelligent agent over time.
 KDD focused on concerns raised by examining realworld applications.
39
Answer Business Questions with these
Techniques
1.
who are the most profitable customers?
2.
Is there really a difference between the profitable
customers and the average customer?
3.
But who really are these customers? Can I
characterize them?
4.
Will some particular new customer be profitable?How
much revenue should I expect this customer to
generate?
40
summary
 Data mining is a craft. As with many crafts, there is a
well-defined process that can help to increase the
likelihood of a successful result.
 We will refer back to the data mining process
repeatedly throughout the book, showing how each
fundamental concept fits in.
41
THE END