Database Reference
In-Depth Information
Analytics classified
In this section, we will focus on learning all the popular analytical techniques that
come under one of the discussed paradigms: descriptive, predictive, and prescriptive
• Classification
• Forecasting or prediction or regression
• Clustering
• Optimization
• Simulations
These analytic techniques can perform either of the two:
• Supervised analysis
• Unsupervised analysis
Supervised analysis is a case where the data is known to us. Client also defines a
specific goal for our analysis and in case of unsupervised analysis, the data might be
known to us, but we usually do not start with a definitive target in mind.
Classification is all about identifying a grouping technique for a given dataset in such
a way that depending on a value of the target attribute, the entire dataset can be qual-
ified to belong to a class. This is one of the techniques used in data mining to identify
the data behavior patterns.
Let's take an example, a marketing manager looking at his customer data wants to
identify if a given customer is helping him make profits and take a decision on if it's
worth spending effort and time on the customer demands. This is commonly referred
to as Total Lifetime Value (TLV).
We have the data and we start by plotting on a graph as shown in the following figure
(the one on the left) not really worrying about what this plotting would mean. On the
y-axis, we have the total money spent (in multiples of hundreds of rupees) and on the
x-axis, we have the number of items purchased. As a next step, we categorize the
data on the graph into good and bad customers, for example. In the following graph,
Search WWH ::

Custom Search