This document outlines the key steps and considerations for writing a dissertation in data analytics, including identifying patterns in data, deriving insights, developing predictive models, and using models to make decisions. It emphasizes that dissertations should apply these analytical activities to address real-world problems or opportunities. The dissertation should demonstrate that the stated research was actually conducted and convincingly report the solutions found. Various research methods, tools, types of data, and analytical project types are also discussed.
2. An analyst cocktail of activities
Identifying data patterns.
Deriving inferences from the data patterns (Insights).
Using the inferences to develop predictive models.
Using predictive models for making reasonable decisions.
Run simulations to investigate scenarios.
Predicting future phenomena or behaviours.
Volume, Variety, Veracity, Velocity
3. From these activities it should be inferred that dissertations in data
analytics should involve these activities in solving pertinent real life
problems or harnessing opportunities.
Researchers should not be tempted to give descriptions or narrations of
these activities in their dissertation.
Writing a dissertation is reporting on the study or research that was
done.
In this light, researchers are expected to carry out the said activities
to address some research question(s) and then explain how that
was done in the dissertation write up.
Thus, in the dissertation the researcher should convince the reader that
the research was actually carried out and a solution founded upon
that particular research.
The work should flow in such a manner that the reader can easily follow
how the methodology was employed in addressing the research
4. A Dissertation in MSCBDA
Data Analytics is about Decision Sciences
Data science Data Engineering
cleans and analyzes data,
answers questions, and
provides metrics to solve
business problems
develops, tests, and maintains data
pipelines and architectures, which the
data scientist uses for analysis.
Does legwork to help the data scientist
provide accurate metrics
5. The Data Science Dissertation
You have explored big data and machine
learning, and how they are used by
organisations of every size.
You have learnt skills and ways of thinking as
you explore data acquisition, preparation,
transformation and modelling.
6. Data Analysis Plan Overview
If the question examines the impact of variable
x on variable y, we are talking about regressions.
If the question seeks associations or
relationships, we are into correlation and chi-
square tests,
If differences are examined, then t-tests and
ANOVA’s are likely the correct test.
7. Finding a good research topic
A dissertation topic doesn’t just appear in your mind
It takes effort and sustained engagement
If you don’t enjoy the topic enough, you’re likely to run into
writer’s block or even writer burnout.
Many students fall into the trap of choosing a topic based on
personal interest alone.This approach is ineffective and will
irritate your advisor.
The best way to avoid this pitfall is to take an objective approach
and brainstorm ideas from a variety of sources.
8. Read example dissertations
You should spend some time reading various
examples of dissertations.This will help you to
know what the expectations are from
different departments.
9. RESEARCH METHODOLOGY
Research methodology is a blue print of how the
researcher carried out the study i.e. it is a system
of methods used scientifically to solve a research
problem.
Technically, research also concerns itself with the
development, examination, verification and
refinement of research methods, procedures,
techniques and tools that form the body of research
methodology.
11. RESEARCH METHODS
Remember research methodology is a system of methods
used scientifically for solving the research problem.
A research method is a technique applied by the researcher to
undertake research.
General research methods available for use include:
1. Experiment
2. Simulation
3. Correlation.
4. Naturalistic Observation.
5. Survey.
6. Case Study.
12. RESEARCHTOOLS/INSTRUMENTS
A research tool or instrument is any means of collecting data or
information necessary for the study.
Traditional research tools include:
Observation forms
Interview schedules
Interview guides
Questionnaires
Data Sets
13. TYPES OF DATATHAT MAY BE
COLLECTED
Data may be grouped into four main types based on methods for
collection i.e. observational, experimental, simulation and derived.
In writing a dissertation, the researcher should not be tempted to
lecture the reader on the various aspects of research
methodology. Rather, they should concentrate on the methods,
procedures and tools that they actually employed indicating:
1. Why they were appropriate
2. How they were used in the current study
3. Their positive and negative contributions
4. How any identified negative contributions were circumvented
14. Pillars of the data science research
methodology
The data science methodology is hinged on
the knowledge discovery (KD)
process;
the Cross Industry Standard Process
for Data Mining (CRISP-DM;
Sampling, Exploring, Modifying,
Modeling and Assessing SEMMA and
Design Science (DS).
Team Data Science Process - TDSP
15.
16. RESEARCHTHE DATA SCIENCE WAY
Data scientists are more concerned with the process used to perform the
analysis.They ask questions such as:
1. How large is the data set?
2. What variables could be included?
3. What hypotheses could be formulated before and after the
analysis?
4. What mechanisms could be used to test the validity of the results
17. RESULTS
In order to have comprehensive results, data scientists design their
research projects to test and re-test their hypothesis in statistically
rigorous ways.These methods include:
Partitioning the data into subsets to test the hypotheses.
Testing hypotheses apparently confirmed in one analysis
against a fresh data set to see if they prove to be predictive.
Use mathematical inference to generalise results to compare
against findings of a specific case.
Using data simulations to create a truly random target set to
compare to genuine datasets
Running comparative visualizations to see results in different
formats.
18. MAJOR KINDS OF DATA ANALYTICS PROJECTS
Descriptive: - Current status
Diagnostic: - Why did it happen (Statistical Analysis)
Predictive: - What will happen (Forecasting)
Prescriptive: - How do we solve it
19. MODELS
❖Mainly data analytics research projects produce models
❖The major activities in these projects therefore entail:
Designing the model
building./developing the model
Evaluating and testing the model
Deploying the model
Getting feedback
20. Some research tools in data analytics
Whatagraph
Xplenty
Zoho Analytics
Juicebox
HubSpot
RapidMiner
R-Programming.
21. When you are done !
Ask yourself the following questions
Is my dissertation title correct ?
Did I give an answer to the research problem ?
What are the opportunities for further research ?