Data Science

learning outcomes of a unit. One credit is equal to 10 hours
of TQT.
ENTRY REQUIREMENTS
For entry onto the OTHM Level 7 Diploma in Data Science qualification, learners must
possess:
● An honours degree in related subject or UK level 6 diploma or an equivalent overseas
qualification
● Mature learners with management experience (learners must check with the delivery
centre regarding this experience prior to registering for the programme)
● Learner must be 21 years old or older at the beginning of the course
English requirements: If a learner is not from a majority English-speaking country must
provide evidence of English language competency. For more information visit English
Language Expectations page on our website www.othm.org.uk.
Alternative professional qualifications with at least three years' relevant work experience in the
public service field may also be considered. This could be in roles in local or national
government, or in non-governmental and inter-governmental organisations, the voluntary and
charitable sector, and private sector roles which support or deliver public services.
PROGRESSION

The OTHM Level 7 Diploma in Data Science enables learners to progress into or within
employment and/or continue their further study.

As this qualification is approved and regulated by Ofqual (Office of the Qualifications and
Examinations Regulation), learners maybe eligible to progress to Master’s top-up at many
universities in the UK and overseas with advanced standing. For more information visit the
University Progressions page on the OTHM website.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
DELIVERY OF OTHM QUALIFICATIONS

OTHM do not specify the mode of delivery for its qualifications, therefore OTHM Centres are
free to deliver this qualification using any mode of delivery that meets the needs of their
Learners. However, OTHM Centres should consider the Learners’ complete learning
experience when designing the delivery of programmes.

OTHM Centres must ensure that the chosen mode of delivery does not unlawfully or unfairly
discriminate, whether directly or indirectly, and that equality of opportunity is promoted. Where
it is reasonable and practicable to do so, it will take steps to address identified inequalities or
barriers that may arise.

Guided Learning Hours (GLH) which are listed in each unit gives the Centres the number of
hours of teacher-supervised or direct study time likely to be required to teach that unit.

The qualification has been designed to take learners on a structured learning pathway. The
sequencing of units is likely to encourage proactive engagement due to the nature of the
subjects and topics therein, whilst also supporting learners to develop the learning and
assessment skills required to be successful at level 7.
CENTRE RESOURCE REQUIREMENTS

Tutor / Assessor Requirements
● Tutors/Assessors must be appropriately qualified and occupationally competent in
the areas in which they are training.
● They must hold a Level 6 qualification or equivalent
● They should hold or be working towards a Level 3 qualification in Assessing
Vocationally Related Achievement such as the OTHM Level 3 Award in Assessing
Vocationally Related Achievement.

Internal Verifier Requirements
● Internal quality assurers or verifiers must be appropriately qualified and
occupationally competent in the areas in which they are moderating.
● They must hold or be working towards a Level 4 Award in the Internal Quality
Assurance of Assessment Processes and Practice and/or a Level 4 Certificate in
Leading the Internal Quality Assurance of Assessment Processes and Practice
such as the OTHM Level 4 Certificate in Leading the Internal Quality Assurance of
Assessment Processes and Practice.
● They must demonstrate that they have undertaken Continued Professional
Development (CPD) activities relating to occupational health and safety or auditing
quality assurance to maintain and update their skills and knowledge within the last
year.

OTHM will request to see copies of relevant qualifications from assessors and verifiers.
ASSESSMENT AND VERIFICATION

The units in this qualification are internally assessed by the centre and externally verified by
OTHM. The qualifications are criterion referenced, based on the achievement of all the
specified learning outcomes.

To achieve a ‘pass’ for a unit, learners must provide evidence to demonstrate that they have
fulfilled all the learning outcomes and meet the standards specified by all assessment criteria.
Judgement that the learners have successfully fulfilled the assessment criteria is made by the
Assessor.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025

The Assessor should provide an audit trail showing how the judgement of the learners’ overall
achievement has been arrived at.

Specific assessment guidance and relevant marking criteria for each unit are made available
in the Assignment Brief document. These are made available to centres immediately after
registration of one or more learners.
OPPORTUNITIES FOR LEARNERS TO PASS

Centres are responsible for managing learners who have not achieved a Pass for the
qualification having completed the assessment. However, OTHM expects at a minimum that
centres must have in place a clear feedback mechanism to learners by which they can
effectively retrain the learner in all the areas required before re-assessing the learner.
RECOGNITION OF PRIOR LEARNING AND ACHIEVEMENT

Recognition of Prior Learning (RPL) is a method of assessment that considers whether
learners can demonstrate that they can meet the assessment requirements for a unit through
knowledge, understanding or skills they already possess and do not need to develop through
a course of learning.

RPL policies and procedures have been developed over time, which has led to the use of a
number of terms to describe the process. Among the most common are:
● Accreditation of Prior Learning (APL)
● Accreditation of Prior Experiential Learning (APEL)
● Accreditation of Prior Achievement (APA)
● Accreditation of Prior Learning and Achievement (APLA).

All evidence must be evaluated with reference to the stipulated learning outcomes and
assessment criteria against the respective unit(s). The assessor must be satisfied that the
evidence produced by the learner meets the assessment standard established by the learning
outcome and its related assessment criteria at that particular level.

Most often RPL will be used for units. It is not acceptable to claim for an entire qualification
through RPL. Where evidence is assessed to be only sufficient to cover one or more learning
outcomes, or to partly meet the need of a learning outcome, then additional assessment
methods should be used to generate sufficient evidence to be able to award the learning
outcome(s) for the whole unit. This may include a combination of units where applicable.

EQUALITY AND DIVERSITY

OTHM provides equality and diversity training to staff and consultants. This makes clear that
staff and consultants must comply with the requirements of the Equality Act 2010, and all other
related equality and diversity legislation, in relation to our qualifications.

We develop and revise our qualifications to avoid, where possible, any feature that might
disadvantage learners because of their age, disability, gender, pregnancy or maternity, race,
religion or belief, and sexual orientation.

If a specific qualification requires a feature that might disadvantage a particular group (e.g. a
legal requirement regarding health and safety in the workplace), we will clarify this explicitly in
the qualification specification.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
UNIT SPECIFICATIONS

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
DATA SCIENCE FOUNDATIONS

Learning Outcomes, Assessment Criteria, and Indicative Content

Learning Outcomes –
the learner will:
Assessment Criteria –
the learner can:
Indicative content
1. Understand the scope of data science
and the roles of data scientists.
1.1 Define the landscape of Data Science.
1.2 Evaluate key topics in Data Science.
1.3 Analyse the role of a Data Scientist in
comparison to other IT roles.

Data Science
Discussion of key topics, their roles, terms, and
definitions in data science, for example:
● Statistics and probability.
● Data science, data mining, data
analytics, and data visualisation.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
● Artificial Intelligence, machine learning,
and deep learning.
● Data-driven approaches and big data.
● Programming, data structures, scientific
computing, and cloud computing.

Role of a Data Scientist
Data scientist roles include:
● Collection and extraction of data from
multiple sources.
● Analysis of data from multiple angles to
produce insights. Looking for trends that
highlight problems or opportunities.
● Mining vast amounts of data for valuable
and actionable insights.
● Effective visualisation of data to highlight
key results.
● Communication of important information
and insights to business and IT leaders
to enable effective decision making,
operational excellence and business
performance.
● Using insights acquired from data
analysis to influence how an organisation
approaches business challenges.
Making data driven recommendations for
business strategy.
2. Understand the impact of big data on
society.
2.1 Define big data.
2.2 Evaluate the impact of big data on users
and organisations for organisational
decision making.
2.3 Critically analyse how Big Data is driving
digital transformation.
Big Data
Big data vs. traditional data management: key
challenges and differences in approach.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
2.4 Critically analyse how big data and
traditional data management differ.
2.5 Evaluate industry-leading tools and
software for analysing and visualising data.
Fundamental characteristics of big data e.g.,
Doug Laney’s Three Vs of Big Data (volume,
velocity, and variety), as well as an extension of
the Vs (variability, veracity, visualisation and
value).
Data-Driven Decision-Making
Advantages of data-driven decision-making:
● Continuous improvement and planning.
● Real-time insights and identifying new
opportunities.
● Cost reduction.
● Aligning decision making with business
strategy.

Challenges for data-driven approaches, for
example:
● Inconsistent and unstandardised data.
● Bias and discrimination inherent in
datasets or sampling approaches.

Value that digital transformation projects can
bring to business and how they achieve this
e.g.:
● Revenue.
● Employee retention.
● Increased productivity.
● Creative performance.
● Brand sentiment.
● Customer satisfaction.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025

Data Science Tools
Industry leading tools and software solutions to
analyse data:
● Programming languages and software
e.g. Python, R, Matlab, Microsoft Excel.
● Data analytics tools: Oracle Analytics,
Qlik Analytics platform, Google Fusion
Tables, Open Refine, Apache
Spark/Hadoop, SAS Sentiment Analysis,
Node XL.
● Data visualisation tools and software e.g.
Power BI, Google Chart, Canvas,
Tableau, Oracle Visual Analyzer, SAS
Visual Analytics, Matplotlib,
Tensorboard.
● Deep learning libraries e.g. Tensorflow,
Pytorch.
● Cloud services e.g. AWS, Microsoft
Azure.
● Database tools e.g., SQL, MySQL.


3. Understand the legal and ethical
responsibilities of data scientists and the
challenges they face.
3.1 Explain the legal and ethical roles,
responsibilities and challenges faced by
data specialists.
3.2 Describe approaches for building ethics
into data science.
3.3 Review the different strategies used by
data specialists to ensure data compliance.
Legal and Ethical Considerations for Data
Scientists
Data protection, informed consent, and privacy
issues for compliance to include:
● Personally identifiable information.
● Sensitive information e.g. protected
health information.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025


Assessment

To achieve a ‘pass’ for this unit, learners must provide evidence to demonstrate that they have fulfilled all the learning outcomes and meet the
standards specified by all assessment criteria.

Learning Outcomes to be met Assessment criteria to be covered Type of assessment Word count (approx. length)
All LO 1 to 3 All AC under LO 1 to 3 Report 3500 words

● General Data Protection Regulation
(GDPR) rights and obligation,
enforcement, and regulatory legal
penalties.
● How legal rules for data compliance
differ globally, challenges faced by global
corporations in collecting and managing
data on individuals.
● Ethical and privacy considerations for
automated and large scale data
collection.

Addressing legal and ethical challenges
Communities and frameworks for good practice
in data science e.g. Data for Good Exchange
(D4GX); Fairness, Accountability and
Transparency in Machine Learning group
(FAT/ML); Data Ethics Framework (gov.uk).
Industry-leading compliance management
software and tools, e.g., Microsoft Compliance
Manager, Amazon Web Services (AWS)
Compliance, IBM DataOps.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
Indicative Reading List
Kelleher, J.D. & Tierney, B. (2018). Data Science (The MIT Press Essential Knowledge Series) (Illustrated edition). The MIT Press.
Kotu, V., & Deshpande, B. (2019). Data Science: Concepts and Practice. Morgan Kaufmann Publishers, an imprint of Elsevier.
Grus, J. (2019). Data Science from Scratch: First Principles with Python (2nd edition). O’Reilly.
Goodfellow, I. & Bengio, Y. & Courville A. (2016) Deep Learning. The MIT Press.
Norvig, P. & Russell, S. (2021). Artificial Intelligence: A Modern Approach (4th edition). Pearson.
Nussbaumer Knaflic C. (2015). Storytelling with Data: A Data Visualization Guide for Business Professionals. Wiley.
Hill, D. G. (2019). Data Protection: Governance, Risk Management, and Compliance. CRC Press.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
PROBABILITY AND STATISTICS FOR DATA ANALYSIS

Learning Outcomes, Assessment Criteria, and Indicative Content

Learning Outcomes –
the learner will:
Assessment Criteria –
the learner can:
Indicative content
1. Understand the fundamentals of
probability and statistics.
1.1 Define probability and statistics, and
explain the difference between them.
1.2 Explain the notion of a probability
distribution and give examples of well-
known distributions.
1.3 Understand the role of stochastic
processes in modelling sequences of
random events.
1.4 Describe key statistics used to describe
sets of events.
Introduction to Probability and Statistics
Probability: the mathematical study of the
likelihood of events to occur. Statistics: the
analysis of the frequency of occurrence of past
events. Understanding the difference between
an underlying and an empirical distribution.
Key concepts in probability and statistics:

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
1.5 Understand the role of hypothesis testing
in using data to answer questions.
1.6 Describe methods for probability
distribution fitting with examples.
● Random variables and observed values.
● Independence and dependence.
● Sampling.
● Cumulative distribution function vs.
probability density function.
● Discrete and continuous probability
distributions.

Common probability distributions – the Bernoulli
distribution, Binomial distribution, and Normal
distribution
Stochastic processes for modelling sequences
of random events over time. Examples of
stochastic processes - Bernoulli process,
Wiener process, Poisson process, Markov
chains, Random Walk.
Statistics and Inference
Essential statistics for probability distributions:
● Mean, median, and mode.
● Variance and standard deviation.
● Understanding what each variable tells
us about a set of data samples and/or
the true distribution.
Key concepts in hypothesis testing: p-values,
null and alternative hypotheses, confidence
intervals.
Approaches to distribution fitting, e.g.:
● Maximum likelihood
● Method of moments

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
● Maximum spacing estimation
● Method of L-moments
2. Be able to use statistics to test
hypotheses on datasets.
2.1 Analyse how inferential statistical analysis
differs from descriptive statistics.
2.2 Evaluate the need for different degrees of
parameterisation in statistical models.
2.3 Explain the distinction between linear,
generalised linear, and non-linear statistical
models.
2.4 Use normal hypothesis tests to evaluate
whether a given claim is true.
2.5 Understand how statistics plays a role in
computational and machine learning models.

Statistics
Descriptive statistics: describing and summarise
the characteristics of a dataset e.g. through the
use of calculating the mean and variance.
Inferential statistics: using the dataset to make
inferences (i.e. drawing conclusions or making
predictions).
Types of statistical models:
● Fully parametric, Semi-parametric, non-
parametric. Advantages and
disadvantages of each (i.e. level of
assumption/bias, level of
variance/uncertainty following model
fitting).
● Linear, Generalised Linear (with a link
function), and Non-Linear statistical
models.

Hypothesis Testing
Hypothesis testing using an assumption of
normality: defining the null and alternative
hypotheses, computing p-values for the claim
using look up tables, verifying whether or not we
reject the null hypothesis based on the p-value.
Statistics and Data Science
The role of statistics in computational data
science and machine learning: e.g. the

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
bias/variance trade-off, parameterisation/over-
parameterisation of models, model complexity,
dataset size.
3. Understand Bayesian statistics. 3.1 Describe the difference between
frequentist and Bayesian views of statistics.
3.1 Explain Bayes' theorem and its importance
and use in statistics.
3.2 Evaluate Bayesian experimental design
with appropriate examples.
3.3 Discuss Markov Chain Monte Carlo
(MCMC) methods and MCMC simulations with
examples.
Frequentist vs. Bayesian viewpoint
Two approaches to estimation of statistical
parameters from samples.
Frequentist: assuming that probability is based
on the frequency of events over time. Leads to
the need to infer via point estimates (estimates
of the probability at a single value on the
distribution support), confidence intervals, and
hypothesis testing (p-values).
Bayesian: assumes that probability is based on
degree of belief in parameters (priors). Inference
updates beliefs
Core difference: consideration of the parameters
as random variables or as fixed.
Comparison of frequentist and Bayesian
approaches:
● Treatment of probability as objective
(frequentist) vs. degree of belief
(Bayesian).
● Parameters are fixed values (frequentist)
vs. random variables (Bayesian).
● Incorporating prior
information/knowledge (Bayesian) vs.
not accounting for this (frequentist).

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
Bayes’ Theorem and Bayesian Inference
Introduction to Bayesian inference and Bayes’
Theorem.
Core components:
● Prior: Encodes beliefs about parameters
before seeing data.
● Likelihood: The probability of the
observed data given the prior on the
parameters.
● Posterior: Beliefs about parameters after
seeing data.
● Marginal likelihood/evidence: The
probability of the observed data.

Using Bayes’ theorem for updating the belief,
the significance of conjugate priors for obtaining
closed-form solutions to the posteriors.
Markov Chain Monte Carlo
A computational Bayesian method for
approximating the posterior distribution
empirically, rather than analytically. Key for
enabling Bayesian inferefence where there is no
closed form solution for the posterior
distribution.
MCMC sampling methods:
● Gibbs sampling
● Metropolis–Hastings algorithm
● Slice sampling

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
● Hamiltonian (or Hybrid) Monte Carlo
(HMC)
MCMC will not work for model comparisons,
where an explicit marginal likelihood (evidence)
is needed.
4. Be able to perform linear statistical
modelling.
4.1 Discuss simple and multiple linear
regression models with examples.
4.2 Discuss and compare linear least squares
methods with examples.
4.3 Define heteroscedasticity, analyse the
need for it, and explain how it is used in the
weighted least squares algorithm.
4.4 Compare traditional statistical and machine
learning approaches to linear regression
Linear Modelling
Distinction between simple linear regression,
multiple regression in terms of number of
independent variables. Multivariate regression:
multiple dependent variables.
Linear least squares methods: Ordinary least
squares (error variances are all the same),
weighted least squares (for heteroscedastic
data, assuming variances of errors differ),
generalised least squares (assumes an arbitrary
covariance for errors).
Regression in statistics vs. machine learning:
● Goal: in machine learning, the goal is to
minimise prediction error using cost
functions and gradient descent vs. in
statistics, where the goal is to form
estimates of statistical parameters.
● Solution: Statistics typically has a closed
form Ordinary Least Squares expression
for linear regression vs. machine
learning methods rely on iterative
optimisation e.g. (stochastic) gradient
descent.
● Analysis and evaluation: Analysis of
results in statistics involves e.g.
significance testing and confidence

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
intervals or residual analysis, whereas in
machine learning it involves comparison
of training/validation/test error.
● Use cases: Machine learning methods
can be useful for higher-dimensional
data, where a closed form statistical
expression is computationally complex.

Assessment

To achieve a ‘pass’ for this unit, learners must provide evidence to demonstrate that they have fulfilled all the learning outcomes and meet the
standards specified by all assessment criteria.

Learning Outcomes to be met Assessment criteria to be covered Type of assessment Word count (approx. length)
All LO 1 to 4 All AC under LO 1 to 4 Report + Source Code 3500 words

Indicative Reading List

Kaptein, M., & den, H. E. van. (2022). Statistics for data scientists: An introduction to probability, statistics, and data analysis. Springer.
Rigdon, S. E., & Fricker, R. D., & Montgomery D. C. (2025). Introduction to Probability and Statistics for Data Science with R. Cambridge
University Press
Nield, T. (2022) Essential Math for Data Science. O’Reilly.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
DATA ANALYSIS AND VISUALISATION

Learning Outcomes, Assessment Criteria, and Indicative Content

Learning Outcomes –
the learner will:
Assessment Criteria –
the learner can:
Indicative content
1. Understand the foundational
principles of data analytics.
1.1 List examples of the three types of data
analytics and evaluate their use in industry.
1.2 Describe the key stages of data analysis
1.3 Demonstrate the ability to prepare datasets
for analysis and visualisation.
Data Analytics
Three types of data analytics:
● Descriptive: Understanding a dataset.
● Predictive: Making predictions for future
outcomes based on current data.
● Prescriptive: Providing recommendations
for future courses of action based on
data.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
Examples:
● Descriptive: data visualisation, cluster
analysis, factor analysis, univariate and
bivariate analysis.
● Predictive: regression analysis, time
series analysis e.g. ARIMA, sentiment
analysis.
● Prescriptive analytics: supply chain
optimisation, pricing strategies, cohort
analysis.

Data Analytics Pipeline
Stages of data analysis:
● Data curation and preprocessing,
including: sampling,
cleaning/transformation (e.g. outlier
identification, missing values treatment,
normalisation), integration of data from
multiple sources, and dimensionality
reduction.
● For predictive and prescriptive purposes:
fitting models to data following
preprocessing.
● Visualising results and communicating
them to stakeholders in a way which is
comprehensible and effective for the
intended audience.

2. Be able to apply prescriptive analytics
to inform decision making.
2.1 Understand how data can be used to
inform decision making.
2.2 Evaluate the advantages of predictive and
prescriptive analytics for organisations and
Predictive and Prescriptive Analytics
Advantages of data analytics for business, e.g.:

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
businesses.
2.3 Understand different approaches to data-
driven decision making.
2.3 Demonstrate the use of prescriptive
analytics using appropriate software or tools.
● Anticipation of the future using predictive
analytics, identifying trends, and forecasting
demand.
● Recommending optimal actions.
● Improving the quality of decision making by
using data to support arguments
(justifiability).
● Establishing the defensibility of decisions
through data-driven decision making in the
context of prescriptive analytics.

Approaches to decision making using data, e.g.:
● Naïve approach: using simple assumptions
or basing predictions on recent history.
● Average approach: basing predictions on the
mean of past values.
● Advanced approaches: times series,
regression, or machine learning for
forecasting.
● Qualitative vs. quantitative approaches to
decision making. Many predictive methods
are quantitative, but qualitative approaches
can be used where some data is missing.

Techniques for prescriptive analytics:
● Classical optimisation.
● Linear and non-linear programming.
● Dynamic programming.
● Simulation to explore how a model behaves
for different decisions and under different
assumptions.
● Decision analysis: systematic evaluation of
alternatives, taking into account probability,
cost, and benefit.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
Using R or Python for forecasting modelling and
associated libraries or functions (e.g. Python’s
SciPy).
3. Be able to visualise and present data
in an audience-appropriate way.
3.1 Understand the need for effective data
visualisation.
3.2 Understand how to tailor the presentation
of data to a given audience.
3.3 Use appropriate tools and software to
visualise data.
3.4 Critically evaluate data visualisation
approaches.
Data Visualisation
Why do we need to visualise data? Examples of
discussion points:
● To enable information to be conveyed more
concisely and rapidly to a broader audience.
● To tailor the presentation of information to
different audiences.
● To present information in a way that guides
insight and understanding.
● Identification of patterns: trends, outliers.
● To form compelling visual arguments to
propose, justify, and defend decisions.

Considerations for data visualisation: clarity and
simplicity, accuracy, functionality, storytelling,
aesthetics, best practices and rules of thumb.
Visualisation approaches for different data
modalities:
● Categorical and qualitative data: bar charts,
pie chart etc.
● Continuous and quantitative data: line charts,
box plots, histograms, scatter plots etc.
● Time series data: line charts, area charts.
● Relationships between features: scatter
plots, heatmaps, bubble charts etc.

Software for data visualisation, e.g.
● Python and associated libraries (Matplotlib,
Seaborn).

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
● R for statistical visualisation (using ggplot2).
● Interactive dashboards, e.g. Tableau, Power
BI, Plotly (Python).
● Excel and Google Sheets for basic
visualisations.

Case studies on data visualisation, analysing
them and critically evaluating their effectiveness
in conveying information in a clear and concise
way in a manner appropriate for the intended
audience.


Assessment

To achieve a ‘pass’ for this unit, learners must provide evidence to demonstrate that they have fulfilled all the learning outcomes and meet the
standards specified by all assessment criteria.

Learning Outcomes to be met Assessment criteria to be covered Type of assessment Word count (approx. length)
All LO 1 to 3 All AC under LO 1 to 3 Report + Source Code 3500 words

Indicative Reading List

Runkler, T. A. (2020). Data analytics: Models and algorithms for Intelligent Data Analysis. Springer Vieweg.
Chen, Y. D. (2022). Pandas for Everyone: Python Data Analysis (2nd edition). Addison-Wesley Professional.
Delen, D. (2019). Prescriptive Analytics: The Final Frontier for Evidence-Based Management and Optimal Decision Making. FT Press.
Meyer, P. I. (2023). The 6 Pillars of Decision Making. Mind Mentor.
Nussbaumer Knaflic, C. (2015). Storytelling with Data: A Data Visualization Guide for Business Professionals. Wiley.
McCandless, D. (2014) Knowledge is Beautiful. William Collins.
McCandless, D. (2012) Information is Beautiful. Collins.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
ADVANCE PREDICTIVE MODELLING

Learning Outcomes, Assessment Criteria, and Indicative Content

Learning Outcomes –
the learner will:
Assessment Criteria –
the learner can:
Indicative content
1. Understand a range of generalised
linear models.
1.1 Understand the distinction between simple
and generalised linear models.
1.2 Identify the key components of a
generalised linear model.
1.3 Describe a range of common generalised
linear models and their associated link
functions.
1.4 Describe the characteristics of dependent
variables that lead to different types of
generalised linear modelling.
Linear and Generalised Linear Models
Key distinction: generalised linear models
include a link function which allows the output to
follow a different distribution to the input.
Components of generalised linear models:
target distribution (from the exponential family),
linear predictor, and link function.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
1.5 Evaluate the difference between
regression, ordinal regression, and multinomial
regression.


Examples of generalised linear models with
associated link functions:
● Linear model: identity link function for
e.g. regression to a normal distribution.
● Log linear model: logarithm of the
response variable varies linearly
(logarithmic link function) i.e. log(p) for
e.g. Poisson regression.
● Logistic models (log odds): logarithm of
the odds varies linearly (logarithmic link
function with odds) i.e. log(p/(1-p)) for
e.g. binomial regression (binary
classification) or multinomial regression
(multiple classes).

Characteristics of dependent variables
Data types and associated generalised linear
models:
● Continuous vs. categorical data: regression
vs. classification.
● Ordinal vs. multinomial data: ordered vs.
unordered categorical data.
● The range of data: infinite, bounded, positive
etc.



2. Be able to implement regression
methods for continuous, ordered and
unordered data.
2.1 Identify appropriate regression modelling
approaches to apply to different types of data.
2.2 Develop simple linear models using
Generalised Linear Modelling
Examples of model types and their use cases:

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
suitable software, e.g. R or Python.
2.3 Explain the notion of maximum entropy
classification in the context of multinomial
logistic regression.
2.4 Develop a range of generalised linear
models for continuous, nominal, and ordinal
data.
2.5 Apply the Poisson regression model and
discuss and address overdispersion and zero
inflation.
● Linear regression: dependent variable is
a linear function of the independent
variables.
● Logistic regression: suitable for a
binomial distribution.
● Poisson regression (for count data):
makes use of the log linear link function.
Approaches for addressing the problems
with Poisson regression:
o Overdispersion (e.g. use negative
binomial regression).
o Zero inflation (e.g. use mixture
models such as zero-inflated
Poisson).
● Multinomial logistic regression
(multiclass classification): for unordered
categorical data with a multinomial logit
link (inverse is the softmax function).
Notion of maximum entropy
classification.
● Ordinal regression: intermediate
between regression and classification –
discrete classes which have an order –
use the ordered logistic link function, or
ordered probit.

Implementing regression for a variety of models
(including simple linear and logistic regression)
using a popular programming language e.g. R
or Python, with appropriate libraries or functions
(e.g. scipy in Python).
3. Be able to develop survival analysis
models.
3.1 Define survival analysis.
3.2 Compare the hazard function to the
survival function and detail how they are
Survival analysis

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
related.
3.3 Describe use cases for survival analysis.
3.4 Describe and implement the Cox
proportional hazards model for survival
analysis.
Survival analysis involves time-to-event data:
analysing how long it takes until an event
occurs.
Hazard function: the rate (risk) of an event
occurring at a time t.
Survival function: the probability of no event
occurring up to a time t.
The survival function is the exponential of the
negative cumulative hazard function. The
cumulative hazard gives you the expected
number of events over a given interval, and the
survival function is the probability of no events
occurring over such an interval.
Use cases for survival analysis: e.g. modelling
time to failure of electrical components, time for
a patient to recover from an illness (for
modelling hospital capacity), time for customer
retention (e.g. cancelling a subscription).
Using Cox proportional hazards for survival
analysis:
● Cox proportional hazards estimates the
hazard-ratios (relative hazards between
groups), which are assumed constant over
time.
● Predicted hazard ratios allow for estimation
of the relative hazard between groups and
how these are affected by the independent
variables.
● Semi-parametric method: explicitly models
effect of independent variables, but not the
baseline hazard.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
● The model is linear in log of the hazard.

4. Be able to use neural networks for
predictive modelling.
4.1 Evaluate appropriate settings for the
application of deep learning methods to
predictive analysis.
4.2 Describe key architectures used for
predictive modelling using neural networks in
different settings.
4.3 Use neural networks to solve predictive
modelling problems.
Neural Predictive Modelling
Neural network architectures:
● Basic and general architectures: fully
connected layers (multi-layer perceptron),
transformers.
● Time series prediction: convolutional
architectures, LSTMs, transformer
architectures, autoregressive models.

Developing and Training Neural Predictive
Models
Training and testing pipeline, loss function
definition, evaluation metric specification,
baseline implementation.
Choosing networks of appropriate architecture
given the dataset and problem setting (e.g. time
series predictive modelling vs. regression vs.
classification).
Choosing the number of parameters/complexity
based on dataset size, and including inductive
biases through choice of architecture (e.g.
number of parameters/using different types of
models e.g. convolutional).

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
Assessment

To achieve a ‘pass’ for this unit, learners must provide evidence to demonstrate that they have fulfilled all the learning outcomes and meet the
standards specified by all assessment criteria.

Learning Outcomes to be met Assessment criteria to be covered Type of assessment Word count (approx. length)
All LO 1 to 4 All AC under LO 1 to 4 Report + Source Code 3500 words

Indicative Reading List

Kuhn, M., & Johnson, K. (2016). Applied predictive modeling. Springer.
Gelman, A. & Hill, J. (2007) Data Analysis Using Regression and Multilevel/Hierarchical Models. Cambridge University Press.
Agresti, A. (2012) Categorical Data Analysis (3rd edition). Wiley.
Agresti, A. (2015) Foundations of Linear and Generalized Linear Models. Wiley.

Chollet, F. (2021) Deep Learning with Python (2nd edition). Manning Publications.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
DATA MINING, MACHINE LEARNING AND ARTIFICIAL INTELLIGENCE

Learning Outcomes, Assessment Criteria, and Indicative Content

Learning Outcomes –
the learner will:
Assessment Criteria –
the learner can:
Indicative content
1. Understand the meaning of Artificial
Intelligence and its role in data science.
1.1 Define Artificial Intelligence.
1.2 Compare the scope of the fields of
machine learning, deep learning, and
Artificial Intelligence as a whole.
1.3 Differentiate between ANI, AGI and ASI.
1.4 Evaluate the impact of deep learning on
the field of data science.
1.5 Describe key technologies used in modern
Artificial Intelligence.
Artificial Intelligence
Artificial Intelligence (AI) is emulation of tasks by
computers traditionally possible only for humans
to perform, e.g. image classification, natural
language processing, automated data
clustering/labelling, reasoning, and planning.
Philosophical debates surrounding ambitions of

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
1.6 Describe a range of real-world applications
of artificial intelligence in disparate
domains.
simulating human intelligence, consciousness
etc.
Appreciating the difference between AI and its
subfields, e.g., symbolic AI, machine Learning,
deep learning and related interdisciplinary
research areas such as robotics. The
connections of AI and its use in diverse fields,
e.g. computer science, mathematics and
statistics, robotics, neuroscience, computer
vision, and linguistics.
Understanding the terms Artificial Narrow
Intelligence (ANI), Artificial General Intelligence
(AGI), and Artificial Super-intelligence (ASI).
Data-driven approaches in AI: e.g. machine
learning, including traditional methods (e.g.
random forests, k-means clustering, support
vector machines), and modern methods in deep
learning (e.g. generative models, large language
models, vision models, deep neural networks).
Key technologies used in modern deep learning:
transformers, diffusion models, reinforcement
learning, and unsupervised learning for learning
in an automatic way from large amounts of
unlabelled data e.g. text on the internet.
Applications of AI
Business and e-commerce, e.g., chatbots,
visual searches, intelligent virtual assistants.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
Engineering, e.g., Computer Aided Design
(CAD), automation in factories.
Healthcare, e.g., care of the elderly, heart beat
analysis, computer-aided interpretation of
medical images, drug discovery.
2. Be able to use machine learning
methods to address real world problems.
2.1 Describe and compare machine learning
and modern deep learning methods for solving
different problems.
2.2 List common tools and software used to
implement machine learning models.
2.3 Describe and implement processes for
preparing data for training and evaluation.
2.4 Use appropriate machine learning methods
to solve real world problems in data science.
2.5 Analyse and explore technical options to
tune and enhance the performance of machine
learning-based systems.
2.6 Use libraries and online resources to
implement existing deep learning models.

Machine Learning Methods
Classical machine learning methods: Linear
regression, logistic regression, decision tree,
Support Vector Machine (SVM), Naïve Bayes,
K-Nearest Neighbor(s) (KNN), k-means,
gradient boosting.
Introduction to neural networks, deep neural
networks, and frontier models (e.g. large
language models).
Common architectures and training methods:
fully connected layers, convolutional layers,
transformers, stochastic gradient descent using
different optimisers e.g. Adam, RMSprop.
Software frameworks and libraries for deep
learning: PyTorch, Numpy, Pandas, Scikit-
Learn, TensorFlow/Keras, JAX.
Tools for machine learning and deep learning,
e.g.: Azure ML, Google Colab, Hugging Face,
AWS for ML.
Practical Implementation of Machine
Learning Methods
Dataset preparation:

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
● Selecting appropriate data/datasets,
addressing issues such as missing
values, class imbalance. Could include
common dataset downloads using e.g.
PyTorch Datasets.
● Evaluating the reliability of data.
● Feature selection and transformation
(e.g. one hot encoding for categorical
features).
● Normalisation.
● Efficient loading during
training/evaluation.

Building neural networks:
● Common software and tools to build
neural networks (e.g. PyTorch).
● Choosing dataset hyperparameters (e.g.
batch size, output range, train-test split.
● Choosing appropriate loss functions and
training techniques for a given dataset
and data modality (e.g. autoencoders,
mean squared error, cross entropy loss,
contrastive learning e.g. Contrastive
Learning Image Pretraining (CLIP)).
● Defining architectures using e.g.
layerwise definitions, e.g. using the
PyTorch nn package, or TensorFlow’s
Keras.
● Evaluating and visualising training
results, comparing to baselines.

Implementing existing deep learning models:

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
● Using libraries to implement popular
models (e.g. Huggingface, Keras,
Pytorch models).
● Using online repositories e.g. GitHub to
download, reproduce, and extend a wide
variety of models from open source
projects.

3. Understand technical, ethical, social
and legal considerations for AI.

3.1 Analyse technical challenges in using
machine learning models for data science at
scale.
3.2 Evaluate the societal benefits of AI.
3.3 Evaluate key ethical and legal challenges
for modern AI.
3.4 Describe best practices for AI model
development, from a technical, ethical, and
legal perspective.
Ethical, Social, and Legal Issues in AI
Technical challenges and limitations of AI, e.g.:
● Hallucinations in Large Language
Models.
● Limited training data.
● High energy expenditure and hardware
requirements for training frontier models.
● Distributed and parallelised training.

Societal benefits of AI, now and in the future,
e.g.:
● Education, access to knowledge, and
tailored tutoring.
● Improvements to health and medicine
and automated drug discovery.
● Addressing environmental challenges
and monitoring climate change.

Ethical considerations challenges e.g.:
● Ethical concerns relating to training/using
large language models trained on large

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
datasets e.g. copyright and ownership of
training data.
● Use of deep learning in recruiting new
employees (automated employee
recommendations/CV scanning) and
potential for bias.
● Harmful content generation and
deepfakes, and their potential negative
impact on individuals and society.
● Adversarial examples and poisoned
training data.
● Bias, interpretability, and alignment in
large language models.
● Potential to widen socio-economic
inequality due to unequal access to
technologies, unemployment caused by
AI.
● Climate change driven by AI’s
environmental footprint.
● Algorithmic quantitative trading and the
impact on global financial markets.

Good practice in AI model development, e.g.:
● Accurate and clear documentation.
● Role of statistic testing and review in
early defect detection.
● Following specific industry standards
(e.g. GDPR).


Assessment

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
To achieve a ‘pass’ for this unit, learners must provide evidence to demonstrate that they have fulfilled all the learning outcomes and meet the
standards specified by all assessment criteria.

Learning Outcomes to be met Assessment criteria to be covered Type of assessment Word count (approx. length)
All LO 1 to 3 All AC under LO 1 to 3 Report + Source Code 3500 words

Indicative Reading List

Deisenroth, M. P., Faisal, A. A., & Ong, C. S. (2020). Mathematics for Machine Learning. Cambridge University Press.
Russell, S. J., & Norvig, P. (2022). Artificial Intelligence: A modern approach. Pearson.
Bishop, C., & Bishop, H. (2023). Deep Learning: Foundations and Concepts. Springer.
Goodfellow, I., & Bengio, Y., & Courville, A. (2023). Deep Learning. Alanna Maldonado.
Coeckelbergh, M. & Gerrard, L. (2020). AI Ethics. MIT Press Essential Knowledge.
Olson, P. (2024). Supremacy: AI, ChatGPT and the Race That Will Change the World. Macmillan Business.

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
ADVANCED COMPUTING RESEARCH METHODS

Learning Outcomes, Assessment Criteria, and Indicative Content

Learning Outcomes –
the learner will:
Assessment Criteria –
the learner can:
Indicative content
1. Be able to evaluate research
approaches in the computing discipline.
1.1 Appraise appropriate research problems in
your chosen area.
1.2 Develop and justify appropriate research
aims and objectives within a defined scope
and timeframe.
1.3 Critically explore, select and justify
research approaches.
Proposing a Research Project
Qualitative and quantitative approaches to
Computing research
The strengths and weaknesses of different
approaches to public sector research.
SMART objectives; terms of reference; rationale
for selection, public sector confidence.
GANTT charts, Key milestones, project goals

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
1.4 Produce a SMART research plan using a
suitable software.

2. Be able to critically review literature
on a relevant research topic.
2.1 Evaluate different literature sources to find
the most appropriate literature for the chosen
research topic.
2.2 Critically analyse different theoretical
approaches to the research problem.
Literature Review
Conceptualisation of the research problem or
hypothesis. The importance of positioning a
research project in the context of existing
knowledge. Significance and means of providing
benchmarks by which data can be judged.
Key theoretical frameworks for research.
Advantages and limitations of qualitative and
quantitative research approaches and methods
3. Be able to design research
methodologies for a computing research
problem.
3.1 Critically evaluate relevant research
methodologies to reflect the research
objectives.
3.2 Design an appropriate methodology in
terms of the research objectives for a defined
population.
3.3 Justify the methodology selected in terms
of the research objectives within the bounds of
agreed ethical guidelines.
3.4 Propose suitable techniques to use with
quantitative and qualitative data collection and
analysis.
Research Methodologies
Research methods e.g., survey, questionnaire,
observations; ways to test sufficiency, reliability
and validity; definitions of data e.g., primary and
secondary sources, qualitative and quantitative;
literature search and review – its credibility, use
and acceptance; ways to reference sources.
Size and sufficiency of data, assessment of the
reliability and validity of information gathered.
4. Be able to develop a research
proposal.
4.1 Create a research question, literature
review, and methodology.
Research Proposal Writing
Report structure e.g., title, acknowledgements,
contents page, introduction, summary of
literature review, research methods used,

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
4.2 Propose techniques for use with
quantitative or qualitative data collection and
analysis.
findings, recommendations, references,
bibliography, appendices e.g., questionnaires,
surveys.
Referencing e.g., Harvard system.

Assessment

To achieve a ‘pass’ for this unit, learners must provide evidence to demonstrate that they have fulfilled all the learning outcomes and meet the
standards specified by all assessment criteria.

Learning Outcomes to be met Assessment criteria to be covered Type of assessment Word count (approx. length)
All LO 1 to 4 All AC under LO 1 to 4 Report 3500 words

Indicative Reading List

Lauro, N. C. (2018). Data Science and Social Research: Epistemology, methods, technology and applications. Springer.

Additional Resources

Mastering Predictive Analytics with R - Second Edition James D. Miller, Rui Miguel Forte Publisher Packt Publication date: August 2017

OTHM LEVEL 7 DIPLOMA IN DATA SCIENCE

SPECIFICATION | OCTOBER 2025
IMPORTANT NOTE

Whilst we make every effort to keep the information contained in programme specification up
to date, some changes to procedures, regulations, fees matter, timetables, etc may occur
during the course of your studies. You should, therefore, recognise that this booklet serves
only as a useful guide to your learning experience.

For updated information please visit our website www.othm.org.uk