Sales Management Customer Relationship Human Resources Business Performance BI & Data Quality IT Tools & Vendors

Sign-in   Register
Establishing 'Making it Happen' as a 'Formal & Predictable' Discipline
   What is KDD- Data Mining? Knowledge Discovery in Databases Process  

ENCYCLOPEDIA→   Enterprise Intelligence  →   -  KDD-Data Mining  →   -  KDD- Data Mining Overview  → 

Knowledge Discovery in Databases Program

KDD - Data Mining program has two streams of requirements. Business/Functional requirements are centered around growth in revenue & profitability and business process optimization. Non-functional requirements like high response time, accuracy, visualization, metadata management and data quality ensure a continued sponsorship of a KDD program.

The prime objective of KDD is to find patterns and rules automatically, based on the data. Data Mining brings together techniques from machine learning (Artificial Intelligence), pattern recognition, statistics, databases and visualization. The emergence of new algorithms especially suited for large volumes of data and facilities to integrate these algorithms into the database engine have added to the attractiveness of Data Mining. The Data Mining technique is used in unearthing associations, segmentation or classification of large data sets, rule discovery, modeling and identifying functional dependencies.

Goals

Typical goals set for the KDD Program relate to the business augmentation (Bottom as well as top line) and process optimization. The overall goals for data mining program can be defined as following.

  • Improve Sales Revenue using the KDD results e.g. by improving Customer Acquisition, Retention and Share of wallet, Customized product packaging increasing sales.
  • Improve Profit using KDD analysis e.g. by controlling operational costs on campaigns, identifying and reducing risks / frauds, configuring profitable products / channels.
  • Optimize / improve business process with minimal human intervention using data mining techniques e.g. workflow / network bottleneck detection, drug formulation.

Objectives

In an organization, the objectives of the KDD program are specific to the business domain can and can be very specific to cases in line of businesses. But the generic objectives can be as following.

  • Unearthing the valuable information hidden in the data. This can be achieved in terms of prediction of unknown or future values of selected metrics.
  • Improve the business opportunities based on the patterns emerged from the data mining results
  • Reduce the cost by identifying the business process gaps
  • Optimize and enhance the business processes

Functional Requirements

The high level business requirements of the organization can be perceived as given in the following table.

Growth in Revenue

  • Divide and Rule: Understand Customer Behavior through demographical (e.g. geographical, socio-economical, professional, educational, personal) and psycho graphical (e.g. hobbies, preferences, privacy norms, event based response) segmentation / classification and improve loyal customer base through Customer Acquisition and Retention for higher Share of wallet.
  • Lateral Growth: Understand market need to provide better products and services including channels. Find cross domain (different Line of Business) opportunities to grow laterally.
  • Cross Sell / Up Sell: Understand Customer / Market and enable cross and up selling through customized product packaging. Use of Channels / Partners for additional revenue sources.

Growth in Profit

  • Control Operational Costs: Controlling operational cost can be achieved through various mechanisms based on the business, but a few generic examples are as follows.
    • Targeted Campaign
    • Optimized Communication
    • Identifying and treating high cost components
  • Profitable Product and Services configurations: Configuring the products and services to reduce cost and gain profit.
  • Managing Risk
    • Credit Risk: Identifying and scoring credit risk for customers in order to maximize profit. Fraud identification (e.g. Credit card frauds like skimming and burst out)
    • Market Risk: Analyzing market risk (Interest Rate based, Foreign Exchange Based, Equity based) and utilizing the knowledge for low / no impact and more profit.
    • Operational Risk: Minimizing operational risk through better and in-time prediction of faults / bottlenecks. Prediction of “just enough” capital allocation for such risks resulting into better asset management.

Business Process Optimization / Enhancement

  • Bottleneck Detection: Identification of bottlenecks / flaws in business processes to augment productivity as well as stakeholder satisfaction. This plays a critical role in managing corporate performance.
  • Process Automation: Identification of processes to be automated finally resulting into lowered operational expenditure.

Non-Functional Requirements

Apart from the business requirements there can be a few non-functional requirements related to the data mining system. Following table gives and indicative list of these non-functional requirements

  • Data Integration and Cleansing: Data mining has a critical dependency on integrated, clean and well-maintained data. Data mining performed on data from a data warehouse is always an ideal situation. But in the absence of a data warehouse, some amount of pre-processing of data would be needed before deploying data mining.
  • Metadata Management: In order to ensure clean and appropriate data is provided to Data Mining process, it is necessary to manage and integrate the enterprise-wide metadata. This aids is improving quality of data through integrated metadata definitions (measures, business rules), which need to be used for data integration and cleansing.
  • Response Time: While performing data mining on large volumes of data and / or complex algorithms, response time becomes a concern. In such scenarios, capable hardware and software techniques need to be positioned to ensure in-time responses, which are critical for business decisions.
  • Visualization: Sometimes, the output of data mining is extremely complex to understand due to the number of entities and steps involved in the computation e.g. output of credit card skimming detection. It is necessary to have the results in interpretable patterns for which based on the business case visualization techniques may need to be developed.
  • Accuracy / Hit Ratio: In order to use the results from data mining, it needs to be guaranteed with positive hit ratio otherwise the program results into irrecoverable operational cost. The positive and high hit ratio or accuracy depends on various factors e.g. sample data, algorithm, model parameterization, test data and time (/ situation) difference between model generation and application.
 

   What is KDD- Data Mining? Knowledge Discovery in Databases Process  
 
 

Was this page helpful?
 
 
More on KDD- Data Mining Overview
What is KDD- Data Mining?
Knowledge Discovery in Databases Process
Data Mining Technology
KDD- Data Mining Issues & Challenges
Knowledge Discovery in Databases Methodology
Data Mining Techniques- Propensity Modeling
Data Mining Techniques- Predictive Modeling
BUY BI & Data Management Vendors & Tools Evaluation Kit
Read more...
BUY largest on-line Data-Quality Management Kit
Read more...
Additional Channels
Principles & Rules
Free Templates
Glossary
Key Performance Indicators

Most Popular Zones with list of pages crossing 25000 hits  →→→ 
Maximising Sales Performance
Sales Leads Generation through Events
Sales Leads Classification and prioritization
Sales Compensation Management
Sales Leads Management System
Sales Leads Generation through advertising
Read more...
  Customer Relationship Management
What is Customer Segmentation?
Customer Segmentation approach
Exit barriers for Customer Retention
Customer Service and Support - Strategic Role
Customer Satisfaction and Retention- Overview
Read more...
  Human Resources & Leadership
Feedback does not mean only negative feedback
People become the way you treat them
Fostering Innovation
Maximize the output first and then the potential
Customer Focus
Read more...
 
 
Business Performance & Planning
Scorecard Health Checklist
Strategic Business Plan
Internal Info Assessment Report
Business Objectives Drill Down
External Info Assessment Report
Read more...
  Business Intelligence & Data Quality
Data Quality Approach Finalization
Data Quality Analysis considerations
Data Warehouse Project Definition
Business Themes+ Data Mart matrix
Foundation & Conformed Dimensions and Facts
Read more...
  IT Vendors & Tools Management
Data Quality through Data Integration Tools
OLAP Scalability
Connectivity and Computing Support
Data Searching and Matching
Collaboration and Administration Support
Read more...