Descriptive Analytics

Descriptive Analytics

Sheik Abdullah A (Thiagarajar College of Engineering, India), Selvakumar S (G. K. M. College of Engineering and Technology, India) and Ramya C (Thiagarajar College of Engineering, India)
Copyright: © 2017 |Pages: 25
DOI: 10.4018/978-1-5225-2148-8.ch006
OnDemand PDF Download:
$30.00
List Price: $37.50

Abstract

Data analytics has becoming one of the challenging platforms across various domains such as telecom, health care, social media and so on. The challenging and most promising task in analytics is the understanding of various patterns in the data. The mechanism of data retrieval and analysis seems to be the promising one in which the algorithms, techniques, way of processing data are in need with the ability to target upon large volumes of data. There are various types of analytical methods such as predictive analytics, descriptive analytics, text analytics, social media analytics and survival analytics. This chapter mainly focuses towards the mechanism of descriptive analytics its types, algorithms and applications. There are various forms of tools and techniques such as association rule mining, sequence rule mining, and data categorization such as hierarchical and non-hierarchical clustering methods with its variants.
Chapter Preview
Top

Types Of Analytics

There are four types of analytics. They are interpreted in the Figure 1.

Figure 1.

­

Descriptive Analytics

Descriptive analytics generates a summary of past data to find useful information and that data is used for future analysis. It aims to describe the customer behavior while buying the products. Descriptive analytics has no target variable in contrast to predictive analytics. Therefore, descriptive analytics is also known as unsupervised learning. There are three types of descriptive analytics. They are as follows,

  • 1.

    Association rule mining

  • 2.

    Sequence rule mining

  • 3.

    Clustering/Segmentation

Top

Association Rule Mining

In data mining, Association rule mining is the most essential and well known technique. The main idea of this technique is towards extracting the exciting relations or common patterns among groups of items in the transactional or relational databases. The buying of individual product when an additional product is bought represents an association rule. It is commonly used in different zones such as classification, telecommunication systems, clustering, market basket analysis, cross-marketing, loss-leader, catalog design, etc. Knowledge can be gained through the rules generated by association rule mining from the various application data.

There are two steps to mine the association rule. They are,

  • Generate Frequent Itemsets in the Transactional Database: Each and every itemsets should take place at least as often as a predefined min_ supp count.

  • Construct an Association Rules using Generated Frequent Itemsets: Association rules should met the predefined min_ supp and min_ confi.

    • Where, min_ supp is the minimum support count.

    • Min_ confi is the minimum confidence.

Classification of Association Rule Mining

  • Based on the level of abstraction

    • o

      Single Level Association Rule: It represents the attributes at exactly one level

    • o

      Multi-Level Association Rule: It refers to the attributes at more than one level

  • Based on the number of dimension

    • o

      Single-Dimensional Association Rule: It represents the attributes at exactly one dimension

    • o

      Multidimensional Association Rule: It refers to the attributes at more than two levels

  • Based on kinds of values

    • o

      Boolean Association Rule: It describes relationships among presence or absence of items

    • o

      Quantitative Association Rule: It describes the relationships among quantitative items

  • Based on the types of forms to extract

    • o

      Frequent Itemset Mining: It extracts the repeated itemsets from the transactional dataset

    • o

      Sequence Pattern Mining: It examines for the repeated subsequence in a sequence dataset

    • o

      Structured Pattern Mining: It examines for the repeated substructures in a structure dataset

  • Based on several extension to association rule mining

    • o

      Restrictions applied

    • o

      Causality analysis

    • o

      Close itemsets

Different methods to mine association rule

  • 1.

    Apriori algorithm

  • 2.

    FP-Growth Algorithm

Complete Chapter List

Search this Book:
Reset