CRISP-Data Mining

CRoss - Industry Standard Process for Data Mining

The need for a standard process

The data mining process must be reliable and repeatable by people with little data mining background. 

Framework for recording experience

          Allows projects to be replicated

Aid to project planning and management
“Comfort factor” for new adopters
           Demonstrates maturity of Data Mining
           Reduces dependency on “stars”

OVERVIEW

Initiative launched in late 1996 by three “veterans” of data mining market.
        Daimler Chrysler (then Daimler-Benz), SPSS (then ISL) , NCR

Developed and refined through series of workshops (from 1997-1999)

Over 300 organization contributed to the process model

Published CRISP-DM 1.0 (1999)

Over 200 members of the CRISP-DM SIG worldwide
    - DM Vendors - SPSS, NCR, IBM, SAS, SGI, Data Distilleries, Syllogic, etc.
    - System Suppliers / consultants - Cap Gemini, ICL Retail, Deloitte & Touche, etc.
    - End Users  - BT, ABB, Lloyds Bank, AirTouch, Experian, etc.

 FEATURES

Non-proprietary
Application/Industry neutral
Tool neutral
Focus on business issues
  -As well as technical analysis
Framework for guidance
Experience base
   -Templates for Analysis

 

  Life cycle: 6 phases

Business Understanding
     Project objectives and requirements understanding, Data mining problem definition
Data Understanding
      Initial data collection and familiarization, Data quality problems identification
Data Preparation
      Table, record and attribute selection, Data transformation and cleaning
Modeling
      Modeling techniques selection and application, Parameters calibration
Evaluation
      Business objectives & issues achievement evaluation
Deployment
      Result model deployment, Repeatable data mining process implementation

 

 

Phase 1. Business Understanding

• Statement of Business Objective
• Statement of Data Mining Objective
• Statement of Success Criteria

    Focuses on understanding the project objectives and requirements from a business perspective, then converting this knowledge into a data mining problem definition and a preliminary plan designed to achieve the objectives

Determine data mining goals
 - a business goal states objectives in business terminology
 - a data mining goal states project objectives in technical terms
  ex) the business goal: “Increase catalog sales to existing customers.”

       a data mining goal: “Predict how many widgets a customer will buy, given their  purchases over the past three years,demographic information (age, salary, city) and             the price of the item.”

Produce project plan
 - describe the intended plan for achieving the data mining goals and the business goals
 - the plan should specify the anticipated set of steps to be performed during the rest of the project including an initial selection of tools and techniques

Phase 2. Data Understanding

• Explore the Data
• Verify the Quality
• Find Outliers

 
    Starts with an initial data collection and proceeds with activities in order to get familiar with the data, to identify data quality problems, to discover first insights into the data or to detect interesting subsets to form hypotheses for hidden information.

Collect initial data
 - acquire within the project the data listed in the project resources
 - includes data loading if necessary for data understanding
 - possibly leads to initial data preparation steps
 - if acquiring multiple data sources, integration is an additional issue, either here or in the later data preparation phase

Describe data
 - examine the “gross” or “surface” properties of the acquired data
 - report on the results

Explore data
 - tackles the data mining questions, which can be addressed using querying, visualization and reporting including:
       distribution of key attributes, results of simple aggregations
       relations between pairs or small numbers of attributes
       properties of significant sub-populations, simple statistical analyses
 - may address directly the data mining goals
 - may contribute to or refine the data description and quality reports
 - may feed into the transformation and other data preparation needed
Verify data quality
 - examine the quality of the data, addressing questions such as:
       “Is the data complete?”, Are there missing values in the data?”

Phase 3. Data Preparation

Takes usually over 90% of the time
        - Collection
        - Assessment
        - Consolidation and Cleaning
        - Data selection
        - Transformations

Covers all activities to construct the final dataset from the initial raw data. Data preparation tasks are likely to be performed multiple times and not in any prescribed order. Tasks include table, record and attribute selection as well as transformation and cleaning of data for modeling tools.

Select data
 - decide on the data to be used for analysis
 - criteria include relevance to the data mining goals, quality and technical constraints such as limits on data volume or data types
 - covers selection of attributes as well as selection of records in a table

Clean data
 - raise the data quality to the level required by the selected analysis techniques
 - may involve selection of clean subsets of the data, the insertion of suitable defaults or more ambitious techniques such as the estimation of missing data by modeling

Construct data
 - constructive data preparation operations such as the production of derived attributes, entire new records or transformed values for existing attributes

Integrate data
  - methods whereby information is combined from multiple tables or records to create new records or values

Format data
 - formatting transformations refer to primarily syntactic modifications made to the data that do not change its meaning, but might be required by the modeling tool

Phase 4. Modeling

Select the modeling technique
        (based upon the data mining objective)
Build model
        (Parameter settings)
Assess model (rank the models)
    Various modeling techniques are selected and applied and their parameters are calibrated to optimal values. Some techniques have specific requirements on the form of data. Therefore, stepping back to the data preparation phase is often necessary.
Select modeling technique
 - select the actual modeling technique that is to be used
  ex) decision tree, neural network
 - if multiple techniques are applied, perform this task for each techniques separately

Generate test design
 - before actually building a model, generate a procedure or mechanism to test the model’s quality and validity
  ex) In classification, it is common to use error rates as quality measures for data mining models. Therefore, typically separate the dataset  into train and test set, build the model on the train set and estimate its quality on the separate test set 

Build model
 - run the modeling tool on the prepared dataset to create one or more models

Assess model
 - interprets the models according to his domain knowledge, the data mining success criteria and the desired test design
 - judges the success of the application of modeling and discovery techniques more technically
 - contacts business analysts and domain experts later in order to discuss the data mining results in the business context
 - only consider models whereas the evaluation phase also takes into account all other results that were produced in the course of the project 

Phase 5. Evaluation

Evaluation of model
    - how well it performed on test data
Methods and criteria
    - depend on model type
Interpretation of model
  - important or not, easy or hard depends on algorithm
    Thoroughly evaluate the model and review the steps executed to construct the model to be certain it properly achieves the business objectives. A key objective is to determine if there is some important business issue that has not been sufficiently considered. At the end of this phase, a decision on the use of the data mining results should be reached

Evaluate results

 - assesses the degree to which the model meets the business objectives
 - seeks to determine if there is some business reason why this model is deficient
 - test the model(s) on test applications in the real application if time and budget constraints permit
 - also assesses other data mining results generated
 - unveil additional challenges, information or hints for future directions

Review process
 - do a more thorough review of the data mining engagement in order to determine if there is any important factor or task that has somehow been overlooked
 - review the quality assurance issues
  ex) “Did we correctly build the model?”

Determine next steps
 - decides how to proceed at this stage
 - decides whether to finish the project and move on to deployment if appropriate or whether to initiate further iterations or set up new data mining projects
 - include analyses of remaining resources and budget that influences the decisions

Phase 6. Deployment

Determine how the results need to be utilized
 Who needs to use them?
 How often do they need to be used

 Deploy Data Mining results by
    Scoring a database, utilizing results as business rules,
    interactive scoring on-line

 The knowledge gained will need to be organized and presented in a way that the customer can use it. However, depending on the requirements, the deployment phase can be as simple as generating a report or as complex as implementing a repeatable data mining process across the enterprise.

Plan deployment
 - in order to deploy the data mining result(s) into the business, takes the evaluation results and concludes a strategy for deployment
 - document the procedure for later deployment

Plan monitoring and maintenance
 - important if the data mining results become part of the day-to-day business and it environment
 - helps to avoid unnecessarily long periods of incorrect usage of data mining results
 - needs a detailed on monitoring process
 - takes into account the specific type of deployment

Produce final report
 - the project leader and his team write up a final report
 - may be only a summary of the project and its experiences
 - may be a final and comprehensive presentation of the data mining result(s)

Review project
 - assess what went right and what went wrong, what was done well and what needs to be improved

 

Summary

CRISP-DM provides a uniform framework for
         - guidelines
         - experience documentation
    CRISP-DM is flexible to account for differences
        - Different business/agency problems
        - Different data