Filter by keyword, subject, or both. Updates live as new model answers are added to our portal.
Time Series Modelling
Time Series Modelling Case Study – Oil Price Forecasting
This individual Time Series Modelling Case Study focuses on analysing and forecasting oil price data using established time-series techniques and an alternative modelling approach. The assessment requires students to work with daily oil price information covering the period from 2024 to 2026 and investigate the underlying patterns, stationarity and forecasting behaviour of the data. The first part of the assignment involves exploratory data analysis and time-series modelling using an ARMA-based approach. Students are required to create appropriate visualisations of the data, perform exploratory analysis and conduct tests for non-stationarity, including relevant stationarity diagnostics such as ADF, ACF and PACF analysis and differencing where required. An ARMA model must then be defined, with suitable model parameters identified using the AIC likelihood approach. The assessment requires the student to examine possible combinations of model parameters, assess model residuals, evaluate model performance using appropriate metrics such as RMSE, and produce forecasts extending 24 months into the future. Confidence intervals must also be included with the forecasts. The second part requires students to investigate an alternative modelling solution for the same oil-price time-series data. Possible approaches discussed in the assignment include models such as LSTM and Prophet. Students are expected to conduct a literature review relating to the selected alternative model, build and apply the model, tune relevant hyperparameters where appropriate, generate 24-month forecasts, create suitable visualisations and calculate appropriate evaluation metrics. The final part of the assessment requires a 6–8 page report describing the modelling process, forecasts, analysis and inferences. The report should explain the reasoning behind the analytical and modelling choices rather than simply presenting numerical results. Students are expected to critically discuss why particular approaches were selected, how the modelling decisions may have influenced the results, how forecasts compare with subsequently collected real data where available, and what improvements could be made in future work. The assessment evaluates both the technical implementation and the quality of the written analysis. The code component assesses completion of the modelling and forecasting tasks, stationarity testing, the alternative solution, code quality and annotation. The report component assesses discussion of the analysis and inferences, comparison of the modelling approaches, clarity of interpretation, report structure, appropriate use of figures and suitable academic references. The submission consists of a report in PDF or Word format, with the code submitted separately or through an appropriate repository.
Read Model Answer →
Artificial Intelligence / Natural Language Processing / Deep Learning
Applying Advanced AI Methods for Analysing Text Documents
This Advanced Artificial Intelligence coursework requires students to implement and evaluate natural language processing, natural language understanding and neural-network techniques for analysing text documents. The assessment uses a supplied social-media dataset containing more than 89,000 posts linked to 947 news headlines, with each post labelled as real or fake through a combination of headline ground truth and majority-vote annotation. Each social-media post is treated as an individual text document for classification and topic-analysis purposes. CMP_6059B_7059B_2025_26_CW1-pre… The first major task focuses on identifying fake text documents. Students must preprocess the text using appropriate NLP techniques, transform documents into numerical feature representations and experiment with alternative preprocessing approaches to determine which performs best. A separate unseen test set must be reserved to evaluate generalisation, while the remaining data is used for training both shallow and deep neural-network classifiers. Students are expected to explain and justify how the data is split. CMP_6059B_7059B_2025_26_CW1-pre… Students first design a multi-layer perceptron (MLP) capable of predicting whether documents are real or fake. The architecture must be justified in terms of input dimensions, number of layers, neuron counts, activation functions and outputs. A second deep-learning neural network must then be developed for the same classification problem, with justification of the chosen network structure, layer types, activations and other configuration decisions. CMP_6059B_7059B_2025_26_CW1-pre… For both networks, students train baseline models and select three hyperparameters considered most important for improving performance. These hyperparameters must be tuned systematically, with visuals prepared to show the experimentation process and resulting performance changes. Appropriate evaluation metrics are then used to compare the trained models. The strongest MLP and deep-learning models must be saved so that they can be loaded and tested on unseen data during the final demonstration without retraining. CMP_6059B_7059B_2025_26_CW1-pre… The second task focuses on topic discovery using NLP and NLU techniques. Students perform syntactic preprocessing such as tokenisation, stop-word removal and lemmatisation or stemming, and experiment with at least two different text-representation approaches. Suggested methods include Bag of Words, TF-IDF, LDA, word vectors and word embeddings. Students must interpret the discovered topics and explain how those topics relate to document content, linked news headlines and class labels. The best topic-discovery model or models must also be saved for live analysis during the demonstration. CMP_6059B_7059B_2025_26_CW1-pre… The assessment is completed through a bench demonstration, supported by a maximum of seven PowerPoint slides. The slides should document the system design, model-improvement process, performance evaluation and discussion of results for both fake-document classification and topic discovery. Students also submit a ZIP file containing only their Python source files. The demonstration lasts up to 15 minutes, consisting of approximately 10 minutes for presentation and technical demonstration followed by 5 minutes for questions and transitions. CMP_6059B_7059B_2025_26_CW1-pre… The marking scheme allocates 45% to fake-document identification, including descriptive analysis, preprocessing, MLP design and deep-learning design; 35% to topic discovery, including preprocessing, model development and interpretation; and 20% to the structure, organisation, professionalism and Q&A quality of the demonstration. CMP_6059B_7059B_2025_26_CW1-pre… Important for the public Reference Library: the brief explicitly states that the use of Large Language Models or generative AI to produce any part of the submission is strictly prohibited, including code, data processing, testing, writing or PowerPoint content. Therefore, this entry should remain only a high-level public description of the assessment and should not be presented as material intended for direct student submission. CMP_6059B_7059B_2025_26_CW1-pre…
Read Model Answer →
Data Science / Time Series Analysis / Machine Learning
Time Series Modelling Case Study: Oil Price Forecasting with ARMA and Alternative Models
This Time Series Modelling Case Study requires students to analyse real-world oil-price time-series data and develop forecasting models capable of predicting future values. The assessment combines traditional statistical time-series techniques with an alternative forecasting approach, requiring students to demonstrate practical modelling skills, critical research engagement and evidence-based interpretation of forecasting results. The coursework is completed individually and contributes 40% of the assessment. assing,,, (1) The assessment is divided into three main parts. Part 1 focuses on developing an ARMA-based forecasting model using daily oil-price data covering approximately 2024 to 2026. Students begin with exploratory data analysis and initial visualisation before testing whether the time series is stationary. Where necessary, appropriate transformations or differencing must be applied to obtain stationarity. Students then define an ARMA model and identify suitable p, d and q parameters using an AIC-based model-selection procedure across the parameter ranges specified in the brief. assing,,, (1) Model adequacy must be assessed using diagnostic analysis. Students inspect residuals, generate additional ACF plots, examine residual distributions and evaluate prediction performance using appropriate metrics such as RMSE. The selected model is then used to forecast oil prices 24 months into the future, with appropriate confidence intervals added to communicate forecast uncertainty. assing,,, (1) Part 2 requires students to research and implement an alternative forecasting approach. Suggested examples include LSTM and Prophet, although another appropriate model may be proposed. Students conduct a literature review supporting the alternative method, build and where relevant hyperparameter-tune the model, generate another 24-month forecast, visualise predictions and confidence intervals, and calculate suitable evaluation metrics. This component is intended to demonstrate independent research and the ability to propose an alternative solution rather than relying only on the conventional ARMA approach. assing,,, (1) Part 3 consists of a 6–8 page technical report explaining the modelling process, forecasting results and resulting inferences. The report should provide a critical analysis rather than simply reproducing numerical outputs. Students are expected to explain why results occurred, justify modelling choices, evaluate how those choices influenced performance, compare forecasts with subsequently observed real data where possible, and construct a coherent narrative supported by plots, images, summary statistics and academic literature. Future improvements to the modelling approach should also be critically discussed. assing,,, (1) Submission consists of both the report and working code. The code may be submitted directly or through an accessible Colab or GitHub repository and must reproduce all models, figures and numerical results presented in the report. The assessment allocates 60% of the marks to code and 40% to the report. Within the coding component, modelling and forecasting completion accounts for 40 marks and code quality and annotation for 20 marks. The report is assessed on analysis and inference, methodological justification, comparison of the two modelling approaches, presentation quality, figures and use of appropriate references. assing,,, (1) Key technical expectations include appropriate testing for stationarity, use of methods such as ADF, ACF, PACF and differencing, systematic model selection, forecasting, evaluation and clear comparison between the traditional ARMA model and the chosen alternative approach. Higher-quality work is expected to interpret what the forecasts mean, identify potential improvements and demonstrate sound technical communication rather than merely reporting model outputs. assing,,, (1) Important for the public Reference Library: the brief explicitly states that students must not use generative AI to write the report, and the rubric indicates that AI text-generation use may result in zero marks for the whole assignment. Therefore, the public entry should remain a high-level description of the assessment rather than material intended for direct submission. assing,,, (1)
Read Model Answer →
Data Science / Artificial Intelligence / Generative Modelling
Generative Modelling Case Study: GANs for Medical Imaging, Cybersecurity and Creative AI
This Generative Modelling Case Study requires students to design, implement and evaluate Generative Adversarial Networks (GANs) across a range of synthetic and real-world applications. The coursework develops both theoretical understanding and practical deep-learning skills, with emphasis on building models, evaluating generated data and critically interpreting model performance. The assessment addresses research understanding, originality, future development and scholarly communication in data science. Generative modelling case study… Part 1 – Building and Understanding GANs from Scratch focuses on fundamental GAN concepts using synthetic two-dimensional data. Students first reproduce a sine-wave GAN from the tutorial and then create a second synthetic distribution using either a 2D spiral, a mixture of Gaussians or a noisy parametric curve. They must modify aspects of the GAN architecture, such as activation functions or network depth, and visually compare generated samples with the original data distribution. Generative modelling case study… Part 2 – Real-World GAN Applications extends the work across three application domains. The first application uses the BloodMNIST subset of MedMNIST to train a DCGAN that generates synthetic blood-cell microscope images. Students explore the dataset, analyse class distributions, train the model, monitor generator and discriminator losses and compare real and generated images using both visual inspection and quantitative measures such as the Fréchet Inception Distance (FID). An optional extension involves implementing a class-conditioned GAN capable of generating images from specific categories. Generative modelling case study… The second application addresses cybersecurity using the CICIDS 2017 intrusion-detection dataset. Students construct a GAN that generates synthetic network-traffic feature vectors rather than images. The model is trained using benign and denial-of-service traffic, and generated samples are compared with real traffic using dimensionality-reduction techniques such as PCA or t-SNE. Students must evaluate how closely the synthetic traffic reflects the distribution of genuine network data. An extension allows analysis of the full CICIDS dataset and evaluation across different attack types. Generative modelling case study… The third application explores Creative AI using the Google QuickDraw pizza category. Students implement another DCGAN to generate artificial pizza sketches, track training behaviour across epochs, and compare generated sketches with genuine examples using both visual inspection and quantitative metrics such as FID. Extension work may examine additional QuickDraw categories and investigate how model performance changes with class and sketch complexity. Generative modelling case study… Submission consists of a 6–8 page report together with working code. The report should explain the analysis undertaken, justify modelling decisions, describe the network architectures, interpret the results and incorporate suitable figures, evaluation metrics and references. The accompanying code must reproduce the figures, models and numerical results reported and must execute successfully when tested. Generative modelling case study… The marking scheme places 60% of the marks on code and 40% on the report. Within the coding component, 40 marks relate to completing the GAN modelling tasks and 20 marks assess code quality, modularity and annotation. The report is assessed on discussion and interpretation of the analysis, justification of architectural choices, results presentation, document quality, figures and appropriate academic references. Generative modelling case study… Important for the public Reference Library: the brief explicitly states that students must not use generative AI to write the report, and the rubric states that AI-generated report text can result in zero marks for the whole assignment. Therefore, use this entry only as a high-level public description of the assessment and do not present generated report content as something students can submit directly. Generative modelling case study… Generative modelling case study…
Read Model Answer →
Data Mining / Data Science
Customer Service Escalation Risk Analytics Using Data Mining and Machine Learning
This advanced Data Mining assessment applies the Knowledge Discovery in Databases (KDD) process to a realistic customer-service risk scenario. Acting as a Data Scientist, students analyse a historical Customer Service Escalation Risk dataset to help an organisation identify early indicators of dissatisfaction and operational bottlenecks that may lead to serious or legal customer escalations. The resulting analysis is intended to support strategic decisions relating to staffing, employee training, customer-journey improvement and escalation prevention. CMP-7023B_Assessement_2 (2) The dataset incorporates customer demographics, account characteristics, communication channels, issue categories, operational measures such as waiting times, transfers and SLA breaches, behavioural indicators including sentiment and response delays, and commercial variables such as monthly fees and contract value. The target variable, escalation_level, contains four categories: No escalation, Minor escalation, Serious escalation and Legal escalation. CMP-7023B_Assessement_2 (2) Students begin with data exploration and visualisation, producing appropriate descriptive statistics and identifying patterns, distributions and potential data-quality concerns. They then perform data cleansing, transformation, feature engineering and preprocessing. Variables that may introduce leakage or unreliable predictions because of their meaning, timing or quality must be critically assessed and justified. CMP-7023B_Assessement_2 (2) The supervised-learning stage requires students to develop, tune and compare predictive models using techniques such as k-nearest neighbours, Decision Trees, Support Vector Machines, ensemble methods or neural networks. Appropriate multiclass evaluation metrics must be used, alongside interpretation of influential variables and model behaviour. CMP-7023B_Assessement_2 (2) The assessment also includes unsupervised learning, requiring comparison of clustering methods such as K-Means and hierarchical clustering after removal of the target variable. Students may apply encoding, normalisation and dimensionality-reduction methods such as PCA or t-SNE and must interpret how the resulting clusters relate to escalation behaviour. CMP-7023B_Assessement_2 (2) Overall, the project assesses independent analytical judgement, modelling justification, comparative evaluation and clear communication of actionable findings for both technical and executive audiences. CMP-7023B_Assessement_2 (2) Overview word count: approximately 340 words. AI-use note: AI tools may only assist with small, specific code snippets. Any AI-generated code must be clearly acknowledged and cited, while the analysis, coding decisions, interpretation and final evaluation must remain the student's own work. CMP-7023B_Assessement_2 (2)
Read Model Answer →
Machine Learning / Cloud Computing
2,000 words
Technical Evaluation and Professional Reflection on a Cloud-Based Machine Learning Fraud Detection Solution
This postgraduate individual assessment critically evaluates a cloud-based machine learning solution for financial fraud detection developed as part of a preceding group project. The scenario concerns a financial services organisation seeking to detect fraudulent transactions in order to reduce financial losses and strengthen customer security. The individual report requires students to examine both the technical quality of the developed solution and their own professional contribution to the project. NUL - LD7187 -Assessment Brief … The first component, Technical Evaluation, accounts for 50% of the assessment and has a suggested allocation of approximately 1,000 words. Students critically analyse the group solution with particular attention to data preprocessing, machine learning model choices, evaluation metrics and technical limitations. The analysis should go beyond description by explaining how specific modelling and design decisions influenced the final outcomes of the fraud-detection system. NUL - LD7187 -Assessment Brief … The second component, Reflection and Professional Issues, also accounts for 50% and is approximately 1,000 words. Students critically reflect on their personal contribution, teamwork experience, challenges encountered and lessons learned during the project. The discussion also addresses wider professional and ethical considerations associated with AI and cloud-based machine learning, including bias, fairness, sustainability and data privacy. NUL - LD7187 -Assessment Brief … The assessment is designed to demonstrate critical understanding of machine learning methods, cloud-computing architectures, practical development of machine learning solutions and awareness of the social, ethical and sustainability implications of AI technologies. A Seminar Activity Tracker must also be included as an appendix to provide evidence of weekly participation and knowledge development. NUL - LD7187 -Assessment Brief … NUL - LD7187 -Assessment Brief … Higher-level performance requires a comprehensive connection between technical decisions and outcomes alongside deep reflection on teamwork, professional development, ethics, fairness and sustainability. NUL - LD7187 -Assessment Brief … Overview word count: approximately 300 words. AI-use note: the brief permits AI for limited support such as grammar improvement, structure, organising ideas and suggestions. The student's main content, analysis and conclusions must remain their own, and any AI use must be declared. NUL - LD7187 -Assessment Brief …
Read Model Answer →
Data Science / Deep Learning
3,000 words
Advanced Research Topics (7PAM2016) — Building GANs from Scratch and Applying Them to Medical Imaging, Network Traffic and Sketch Generation
This Masters-level assessment asks for a complete generative adversarial network study, delivered as an annotated code submission carrying sixty per cent of the marks and a six-to-eight page technical report carrying the remaining forty. The work spans four separate GAN implementations, moving from a controlled synthetic setting into three contrasting real-world application domains. Part one builds a GAN from scratch in PyTorch on synthetic two-dimensional data. The tutorial sine-wave generator is reproduced first as a baseline, then a new distribution is modelled — a noisy parametric curve of the form y = sin(2x) + 0.3cos(5x) with an additive noise term — before the architecture itself is varied. Activation functions and layer depth are altered systematically and the resulting sample distributions plotted against the originals, so the effect of each architectural choice on convergence and sample fidelity can be seen rather than asserted. Part two applies the same principles at scale across three domains. The medical strand trains a DCGAN on the OCTMNIST subset of MedMNIST, generating synthetic optical coherence tomography retinal images, tracking generator and discriminator losses across training, and evaluating output both visually and quantitatively using Fréchet Inception Distance. A conditional GAN extension conditions the generator on class label so that images for a chosen retinal pathology can be produced on demand. The cybersecurity strand shifts from images to feature vectors, using preprocessed CICIDS 2017 network intrusion data. Benign and DoS traffic is combined and explored for class balance, a GAN is built to synthesise tabular feature vectors rather than pixels, and real against generated distributions are compared through PCA and t-SNE projections, with a discussion of how well the model generalises across attack types. The creative strand trains a DCGAN on the QuickDraw 'birthday cake' sketch category, tracking visual outputs epoch by epoch and benchmarking generated sketches against real ones, with an extension covering additional categories of differing sketch complexity. The accompanying report explains the analysis steps and the reasoning behind each architectural decision rather than restating textbook definitions of the method. It gives brief descriptions of the models used, presents generated samples and loss curves as figures, interprets the evaluation metrics, and reflects honestly on failure modes — training instability, mode collapse, and the visible flaws in synthetic output that determine whether such data is fit for downstream use. The code is written as reusable functions, commented for a reader other than its author, and reproduces every figure and numerical value quoted in the report.
Read Model Answer →
Artificial Intelligence
2,000 words
End-to-End Applied AI Development — Comparative Machine Learning and Neural Network Modelling on a Public Dataset
This assessment runs a complete applied AI development cycle end to end: problem definition, dataset selection, preprocessing, model building, optimisation, evaluation and critical reflection. Students identify a real-world problem themselves, formulate a research question from it, and source a suitable dataset from a recognised public repository such as UCI, Kaggle, Data.gov or OpenML. Dataset choice carries more weight than students expect. It must be genuinely suitable for supervised learning, complex enough to make preprocessing and feature engineering meaningful, and — critically — structured so that a traditional machine learning approach and a deep learning approach can be sensibly compared on it. A dataset too small or too clean makes the neural network component pointless; one too large or too noisy makes the whole pipeline unfinishable within the page limit. The source must be referenced and the choice explicitly justified against the research problem. The modelling requirement is fixed: at least two supervised machine learning models, plus one artificial neural network built in a mainstream deep learning framework, all trained and tested. The comparison between them is the analytical core of the work. Reporting that the neural network scored higher is not an answer; explaining why, in terms of the data's structure and each model's inductive assumptions, is. Marks are distributed across problem framing, the traditional models, the deep learning model, evaluation and critical analysis including responsible AI considerations, and academic communication. That responsible AI component is easy to overlook and is not decorative — it asks what the model's limitations mean for anyone who might rely on it. Presentation requirements are specific. The report is page-limited rather than purely word-limited, and every plot must be described in the text while also being legible enough to communicate on its own — a common failure is dense default library output pasted in without axis labels or scale. The implementation is documented in a notebook combining markdown and code cells so the development process is visible, not just the final result, and submissions typically include the cleaned dataset alongside the code. The strongest submissions treat the notebook and the report as one argument. Weaker ones produce a working notebook and then write a report that describes it, rather than a report that uses it as evidence. Our support on assessments of this type is guidance-based. Typical areas of help include: advising on whether a candidate dataset can actually support the required model comparison, explaining how to justify preprocessing decisions, clarifying which evaluation metrics suit which problem type and why accuracy alone is often misleading, showing how to structure a critical limitations and responsible AI discussion, checking Harvard referencing, and reviewing a student's own draft against the published marking criteria.
Read Model Answer →