Establishing 'Making it Happen' as a 'Formal & Predictable' Discipline

Execution-MiH ENCYCLOPEDIA → Enterprise Intelligence → SECTION - KDD-Data Mining →

CHAPTER -

KDD- Data Mining Overview

Knowledge Discovery in Database is next level of evolution beyond data query and analytics. It moves from 'what has happened'/'why it has happened' to 'what is likely to happen?'. In tradition analytics, you have to ask right questions to get the right answers. In KDD- Data Mining, it helps you to ask right questions. KDD-Data mining moves from 'retroactive' to 'pro-active'.

Topics

What is KDD- Data Mining?

Data Mining term is used interchangeably with KDD. In reality, it is one of the steps in the whole process of knowledge discovery in databases. Data Mining needs a well defined business case and a diligent data preparation and has to be followed with a detailed evaluation of the discovery results.

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.

Knowledge Discovery in Databases Process

Knowledge Discovery in Databases comprises four key stages in an iterative flow- Business Case Definition, Data Preparation, Data Mining and Evaluation. Data Mining has no value on a stand-alone basis. Its success depends on how well you define the problem and on the level of diligence in data preparation.

Data Mining Technology

Data Mining architecture has three layers- Database Layer with sub-layers of database & metadata, application layer performing data management & algorithms and front-end layer for administration, input parameter settings and results display/visualization.

KDD- Data Mining Issues & Challenges

Key issues around KDD-Data Mining are around limited information, noisy & missing data, level of uncertainty and dynamically & fast-changing data reference.

Knowledge Discovery in Databases Methodology

Knowledge Discovery in Databases (KDD) methodology comprises five phases- Definition (vision/ business case/tools), specification (requirements/techniques/data analysis), design (data preparation/business case model), build and deploy.

Data Mining Techniques- Propensity Modeling

Data Mining Propensity modeling works on discovering a natural inclination or tendency across the variables. This group of techniques includes Cluster Analysis, Association Analysis and Sequential/Temporal patterns

Data Mining Techniques- Predictive Modeling

Predictive modeling is used to create a model of future or expected behavior. This group of data mining techniques includes Induction, Classification, Regression and Neural Networks.