Data Analysis Concepts Revision

Question 1

State the five stages of creating structured data

Answer 1

1) Document scanning
2) Text recognition
3) Character encoding
4) Parsing
5) Migration to database

Question 2

Describe ‘serial’ file organisation

Answer 2

Data that is organised based on the order of their creation. As a result, data is unorganised, or at best is in chronological order.

Question 3

Describe ‘sequential’ file organisation

Answer 3

Data is sorted by a key field, often a primary key.

Question 4

State and briefly explain four ways to achieve good quality data

Answer 4

1) People and skills - with appropriate knowledge and competency in their role
2) Governance and Leadership - policies and procedures in place to support the process
3) Systems and processes - systems to support validation and verification
4) Data security - Ensuring data collected is secure and only used by authorised individuals

Question 5

What does CUVCATR stand for in relation to Data Quality

Answer 5

C - Completness
U - Uniqueness
V - Validity
C - Consistency
A - Accuracy
T - Timeliness
R - Relevance

Question 6

What is the difference between Personal and Identifying information with relation to GDPR

Answer 6

Personal - any record relating to yourself (i.e doctor records)
Identifying - information that can be used to identify from others in a dataset

Question 7

Define ‘data lineage’

Answer 7

Includes the origin of the data, what happens to it and where it moves over time

Question 8

Define ‘Interpolation’

Answer 8

The creation of new estimated data points based on pre-existing data points

Question 9

State and explain the three forms of interpolation

Answer 9

Linear - The simplest form that makes fewest assumptions about the data
Polynomial - Captures non-linear patterns
Nearest Neighbour - Does not generate new values, replicates the nearest existing values

Question 10

State the difference between the Null and Alternative hypothesis’

Answer 10

Null hypothesis states that whatever relationship you are studying is not due to a real effect but observed only because of random sampling

Alternative hypothesis states that the effect/relationship you are measuring/observing is due to a real phenomena.

Question 11

Define ‘Data Architecture’

Answer 11

Data architecture is collective term describing the systems, policies, rules and standards that aim to standardise the way data is collected, handled, stored and transmitted

Question 12

State four advantages of using a data architecture

Answer 12

1) Operations on data are done in the same/similar ways
2) Upgrades/maintenance to software or hardware are simplified
3) Accessing and performing operations on data is made easier
4) Encourages people to think of the wider context in which their application/systems live

Question 13

Define ‘Domain Knowledge’

Answer 13

Knowledge of a specific industry and business

Question 14

Define ‘Descriptive Analysis’

Answer 14

Analysis that shows ‘what has happened?’. Often involves summary stats i.e mean, count, sum.

Question 15

Define ‘Predictive Analysis’

Answer 15

Helps project trends and patterns into the future.

Question 16

Define ‘Decision Analysis’

Answer 16

Helps understand the ‘map’ of possibility from a decision. Uses known outcomes, risks and machine learning to estimate a likely outcome.

Question 17

Define ‘Prescriptive Analysis’

Answer 17

Helps understand the best course of action of possibility from a decision. Similar to ‘decision’ analysis but are scored by a set of criteria.

Question 18

Which of the following would you use to gain consensus;

Focus Groups
Interviews
Shadowing
Surveys
Studying Documents

Answer 18

Focus Groups

Question 19

State three issues that can occur from requirements elicitation

Answer 19

1) Problems with scope - boundaries of systems are ill-defined
2) Problems with understanding - Stakeholders aren’t sure of what is needed
3) problems of volatility - Requirements change over time

Question 20

State and briefly describe the four types of database maintenance activity

Answer 20

1) Log File Maintenance - Updating and deleting old logs, and checking access for unusual activity
2) Data Compaction - Removing old/redundant data and replacing with metadata
3) Integrity Check - Checking for vulnerable pieces of data with insufficient backups.
4) Data Warehousing - Bringing structured data together into the database

Question 21

State five things that make a hypothesis ‘good’

Answer 21

1) MUST be testable and be written in non-ambiguous language
2) MUST at least partly answer the problem statement
3) MUST be used to make at least one clear prediction
4) MUST be based on reliable and relevant information
5) MUST contain dependent variable (what you measure) and an independent variable (something that is controlled or changed)

Question 22

What are the advantages of using both structured and unstructured data

Answer 22

• Faster decision making
• More relevant decisions
• Fewer wasted man hours
• More opportunities spotted
• Rich data sources available

Question 23

What does IPR stand for when referring to data protection rules

Answer 23

Intellectual Property Rights

Question 24

What does ‘CIA’ mean when referring to a model designed to guide information security policies

Answer 24

Confidentiality, Integrity, Availability

Question 25

How to overcome problems with requirements eliciation

Answer 25

Use visualisations to help people understand
Consistent language
Guidelines
Consistent use of templates
Document dependencies between requirements
Analysis of change - route cause analysis

Question 26

Order the following in the order that they would occur as part of an ETL;

Session
Mapping
Source
Target
Workflow

Answer 26

1) Source
2) Target
3) Mapping
4) Source
5) Workflow

Question 27

Define CAP theorem and explain the three elements

Answer 27

The theory that a distributed system can delivery only two of three desired characteristics;

Consistency - All clients see the data at the same time
Availability - Any client making a request for data gets a response
Partition Tolerance - The cluster must continue to work despite any number of communication breakdowns.