Reference no: EM134042469

Assessment:

Assessment type: Group Research and Practical assignment

Assessment purpose: The purpose of this assignment is to assess the students' knowledge about different Artificial Intelligence models for data analysis and prediction, achieving hidden knowledge/insights. By implementing five different AI models for data analysis and prediction, students will gain a solid understanding of AI models. This assessment contributes to learning outcomes a, b, c, d.

Assessment topic: Implementation of AI models on a benchmark dataset

Task Details

Ransomware is a sophisticated type of malware that blocks access to a victim's systems or data until a ransom is paid. In 2026, it remains one of the most widespread cyber threats, increasingly shifting from simple encryption to multi-layered extortion tactics.

Common Types of Ransomware (2026)
Crypto Ransomware: The most prevalent form; it uses advanced encryption algorithms (e.g., AES, RSA) to lock files like documents and databases.
Locker Ransomware: Completely locks users out of their entire device or operating system, often displaying a full-screen ransom demand.
Leakware/Doxware: Steals sensitive data and threatens to release it publicly if payment is not made, regardless of whether the files are encrypted.
Wiper Malware: Destroys data permanently, sometimes disguised as ransomware even if the victim pays.

In this project, you work in a group of 3 or 4 students. You need to construct 5 complete processes using 5 different models to detect a specific ransomware. You can select any specific ransomware such as Crypto Ransomware.

Key Tasks

You need to complete the following key tasks:

Dataset Selection. You need to find a proper dataset on the internet to train your models.

Models. You are required to use 5 different models, and one model must be a neural network model.

Processes. You need to construct 5 different processes to detect a specific ransomware of your choice, and use a dataset to train your models.

You are required to justify all the models by explaining why you choose those models and their strengths and weaknesses. For all the models, you aim to achieve good overall performance for each model by using the following techniques:

Normalisation. About feature (attribute) normalisation, you may read this article.
Parameter tuning. Each model has a few parameters, and those parameters have significant impacts on the model performance. You need to tune those parameters in a systematic way to achieve the best performance.
Attribute selection. You need to select the best subset of attributes to achieve the best performance.
Techniques for handling data imbalance if you have an imbalanced dataset.

Report Structure

You are required to use the following report structure that addresses the marking rubric criteria:

Introduction. Introduces the case study and the objective.
Data understanding. Visualize and explain the data. What preparation methods are required? What features play an important role in your prediction model?

Model selection, implementation and evaluation: Explain in detail why you select these models, how you implement them, how you tune the parameters in order to get better performance, and what metrics you use to measure the performance of a model. You also need to compare the models against each other, and use a single table to show the comparison results. Include the key screenshots of your implementation and analysis in the right place in this section.

Insights: Discuss the output that you receive for the implemented models. Explain the knowledge and insight about the dataset.

Conclusion: Conclude the model implementation and the results you receive from them. Highlight the key points in the insights.

References. All sources used in this assignment must be properly cited using a specific referencing style, e.g., Harvard. Failure to do so will be treated as plagiarism. In-text citations are essential.

Submit all your rapid miner files and your report in a word file on Moodle.

Marking Rubric for Assessment 4

Value: 35%

Rapid Miner Source code - 2%
Fail (0-49%): Performs less than 50% of tasks. Structure not clear.
Pass (50-64%): Performs 50%-64% of tasks. Structure lacking in some parts.
Credit (65-74%): Performs 65%-74% of tasks. Structure not totally clear.
Distinction (75-84%): Performs 75%-84% of tasks. Well structured.
High Distinction (85-100%): Performs more than 85% of tasks. Very clearly structured.

Data Understanding and Preparation - 2%
Unsatisfactory: Unsatisfactory data preparation.
Satisfactory: Satisfactory data preparation but no filtering, no fair distribution of samples.
Good: Good data preparation but lack of filtering.
Excellent: Excellent data preparation but not proper set of attributes are selected or there are outliers.
Exceptional: Exceptional data preparation. Variety of feature selections are tested. Exceptional feature extraction is done. Selected features used combinations of existing and extracted features.

Model Implementation, performance analysis and comparisons - 9%
Unsatisfactory: Unsatisfactory intelligent models.
Satisfactory: Satisfactory intelligent models implemented but the accuracy is poor or there is no proper objective.
Effective: Effective intelligent models with fair accuracy. The parameters could be better adjusted.
Excellent: Excellent intelligent models with good accuracy. Normalisation, attribute selection and parameter tuning properly implemented, and almost complete comparisons. Still there is a chance of better results.
Exceptional: Exceptional intelligent models with excellent outcome. Normalisation, attribute selection and parameter tuning well implemented, and complete comparisons.

Draw a conclusion based on the insights and analysis - 2%
No conclusion: No conclusion presented and no visualization.
Satisfactory: Conclusion based on some argument and satisfactory visualization.
Good: Conclusion based on sound argument and effective visualization.
Excellent: Conclusion based on an extensive argument and excellent visualization.
Exceptional: Conclusion based on an extensive and compelling argument and exceptional visualization.

Individual Contribution - 10%
Unsatisfactory: As a member of a group, you have had unsatisfactory contribution or no contribution.
Satisfactory: As a member of a group, you have had fair minimum contribution in each of the above marking criteria. It should be clearly stated in the report that what contributions the member has had.
Good: As a member of a group, you have had fair contributions in each of the above marking criteria. It should be clearly stated in the report that what contributions the member has had.
Excellent: As a member of a group, you have had good contributions in each of the above marking criteria. It should be clearly stated in the report that what contributions the member has had.
Exceptional: As a member of a group, you have had excellent contributions in each of the above marking criteria. It should be clearly stated in the report that what contributions the member has had.

Group Demonstration - 5%
Unsatisfactory: Very little subject matter knowledge shown.
Satisfactory: Adequate subject matter knowledge shown.
Good: Sound subject matter knowledge shown.
Excellent: High level of subject matter knowledge shown.
Exceptional: High level of subject matter knowledge and synthesis with theory shown.

Individual Demonstration - 5%
Unsatisfactory: Very little subject matter knowledge shown.
Satisfactory: Adequate subject matter knowledge shown.
Good: Sound subject matter knowledge shown.
Excellent: High level of subject matter knowledge shown.
Exceptional: High level of subject matter knowledge and synthesis with theory shown.

Member 1

Primary Responsibilities:
Dataset + Data Understanding & Preparation + 1 Model

Rubric Criteria Covered:
Data Understanding and Preparation (2%) + part of Model implementation (≈2%) + RapidMiner source code contribution

Specific Deliverables:

Find & justify the ransomware dataset
Full data understanding, visualisation, cleaning, normalisation, attribute selection, imbalance handling
Implement & tune Model 1 (e.g. Decision Tree or Random Forest)
Write Data Understanding section of the report
Contribute RapidMiner process file for Model 1

Note: Need only Member 1 Task