Each student in this course is asked to do an individual project consisting of a prototype machine learning application, a term paper documenting the process of developing your application, and an optional presentation if it can be arranged.

The development of your project should include, and your term paper should document, at least the following steps:

• Application domain knowledge analysis: conduct a through (or more reasonably, sufficient) investigation to understand the domain knowledge in your chosen application field. Select one problem that can be solved by building a predictive model.
• Data collection: identify and collect as many relevant data needed by your application as you can. You can use an existing dataset, or collect and assemble your own data from various sources.
  The following list shows a few sources that either provide datasets suitable for machine learning tasks, or data in their raw formats:
  • Kaggle Datasets
  • UC Irvine: UC Irvine Machine Learning Repository
  • OpenML Datasets
  • Open Government Data Portal by Government of Canada
  • British Columbia Data Catalogue
  • BC Baccalaureate Graduates Survey Reports
    (Unfortunately, these reports are in PDF format, so the data isn't readily readable by machines.)
  If you used a readily available dataset for your application, you need to provide the accessible source of the dataset. If you assembled your own dataset, you need to attach the actual dataset in the analytics base table format.
• Data exploration: explore the dataset you assembled in the previous step, justify your attribute selection, and report your exploration result (similar to what's required in Assignment 1).
• Data pre-processing: perform and explain any necessary pre-processing of your data so that they are suitable for your chosen predictive model building method.
• Model building: build a predictive model using a subset of your assembled data. The rest of the data are saved to evaluate your model. Whether you used your own machine learning program or a machine learning platform to build your model, you need to explain the algorithm(s) you choose to use. If you used your own program, attach your source code. If you used a platform, introduce the platform so that your model building process can be replicated.
• Model evaluation: test the predictive model you built in the previous step, explain your test method, present and interpret the test result.
• Model deployment: explain how your model can be used to solve the problem you described in section "Application domain knowledge analysis".

If you need any help to determine whether your project idea works, please send me your application topic and your intended scope, and any preliminary work you've done so far, preferably by the end of October.

If it can be arranged, you will be asked to do a 10-minute presentation to introduce and demonstrate your project during the last week's lab.

The submit deadline of your project is tentatively set at 17:00, 11 December 2026, Friday.

You will receive one mark for your project. Your project will be marked based on the proper selection of the dataset and machine learning algorithms for the prototype application, the clarity and logical coherence of explaining your development process in your term paper and in your presentation if there is one, the depth of your understanding of the machine learning process in general demonstrated in your application and in your term paper, the accuracy of the information conveyed in your project, and the novelty in your topic selection.