CMS 3503 Machine Learning 

Assessment Requirement 2021-2022 

The assessment is to apply machine learning algorithms of your choice to analyse a real-world benchmark problem. Three concrete application areas with benchmark datasets are listed in the Appendix; if you wish to work on another domain, you must consult with your module tutor first. The assessment is 100% coursework and is a portfolio consisting of the following: 

1. Investigation: You should produce a scholarly report concerning the use of the machine learning methods in the application area you’ve selected. You should evaluate both the appropriateness and the readiness of a set of ML techniques to a problem class. This weights 40% of the overall marks.  

2. Development Task: You should then propose a software solution to the practical problem using machine learning method of your choice, which is to be documented in an evaluative report. This weights 60% of the overall marks, and will be included in your final report.  

The report must be submitted electronically. The specific requirement for each component is as follows:

• o Literature review – 1200 words 

? An introduction to the problem domain 

? A critical review of the associated literature 

? The findings or summary of the reviewed literature 

o Solution Reporting – 1800 words 

? A description of the planned research, methodology and evaluation methods 

? A description of the activities undertaken (e.g. any implementation &/or design of experiments) 

? The findings of the work 

? Conclusions and further work 

Please note the specified word count does not include references. And you must explicitly annotate the word count in your report. 

• In order to mark from a holistic perspective and to promote those with potentials for publications, both assignments should be arranged in one piece of writing so
• o you must submit one complete report that covers both Investigation and Development tasks to the (02-1) Development Task portal, which will be marked; 

We require you to use a standard template for your report and have selected that of the IEEE conferences.  This will also make the transition from your report to a possible conference paper or a book chapter easier (in case of enough contributions to knowledge). The templates are in MS word or Latex and can be acquired from: 

• https://www.ieee.org/conferences/publishing/templates.html

Please note: You are required to choose one of the benchmark data sets below for your investigation and development tasktask4. Heart failure clinical records Data Set 

This dataset contains the medical records of 299 patients who had heart failure, collected during their follow-up period, where each patient profile has 13 clinical features. Data source at UCI: https://archive.ics.uci.edu/ml/datasets/Heart+failure+clinical+records 

Reference: Davide Chicco, Giuseppe Jurman: “Machine learning can predict survival of patients with heart failure from serum creatinine and ejection fraction alone”. BMC Medical Informatics and Decision Making 20, 16 (2020)