Business Analytics

Question 1

The concept of analytics

Answer 1

A broader term which describes the various methods of analysis used to understand a particular phenomenon

It captures the different aspects of business problem solving

Question 2

Definition

Answer 2

Burdened analytics refers to all the methods and techniques that are used by an organisation to measure performance

Contain statistical methods that can be applied to a specific project, process or product

They can be used to address problems within the entire company

Performed in order to indenting weaknesses in existing processes and highlight meaningful data that will help an organisation prepare for future growth and challenges

Question 3

What kind of problems do we want to solve

Answer 3

Companies who have analytical maturity (understanding of goals) analytics strategy builds on SMART goals

S: specific (certain area for improvement
M: measurable (by defining specific targets to measure)
A:assignable (who is going to solve it?)
R:realistic (what should be expected subject to a time plan)
T:time related (when are the results going to be delivered)

Question 4

Types of analytics

Answer 4

Exploratory : identifying variables measures and type of data
-identification of coding errors
-data quality issues
Data completeness

Explanatory : extract some explanation as to what affects these variables (e.g. What is the relationship between sales and time)

Predictive: build a model and validated its relevance

Prescriptive: assess the economic significance of the model and implement it In a model lifecycle

Question 5

4 Vs and explanatory analytics

Answer 5

Volume- size: how many observations, are they all uniform ? How many are missing ?

Velocity- time: how fast are the data coming? How often does the database change

Variety - how many types of data do we have ? Are they all structured?

Veracity - accuracy: how accurate are the data ? (E.g. Age in years or through DOB

Question 6

Predictive analytics

Answer 6

Two main approaches

Value protection- (in case of a continuous variable ) of the actual value with some confidence interval

Classification-(in the case of finding out the class of this variable) 
-nominal :true or false 
Ordinal : a rank for this variable or a score
-probabilistic : 20% likelihood for class A

Question 7

Challenges in prescriptive analytics

Answer 7

Optimisation of one measure in expense of the other (internal cannibalisation)

Godhards law: when a measure becomes a target, it ceases to be a good measure

Data silos: when different stakeholders have to be involved and don’t want to share data with other department

Cost of implementation out performs the pay off