MAT013 Group Coursework

Deadline: May 4, 2017

Instructions

The outputs of this coursework will be:

• A 20 minute group presentation/demonstration to be given on 1st, 2nd, 3rd or 4th May 2017 followed by 10 minutes for discussion. The most recommended date is May 2, 2017. However, a group can choose another date due to unavailability on May 2.
• All relevant files (code, presentation, notes, websites, demo materials, etc) should be passed to Andrey Pepelyshev on or before May 04, 2017.

Marking criteria:

• Difficulty: [30]
• Accuracy: [30]
• Originality: [20]
• Presentation/demonstration: [20]

Coursework

As a group you are required to present how to solve a particular scientific problem using R. You should use aspects of R that are not given in the notes. The presentation should be viewed as a teaching presentation. You should follow the following strategy:

• Choose a scientific problem. Some examples of statistical problems are:
  • Logistic/Binary/Poisson regression
  • Classification/Clustering
  • Discrimination analysis
  • Multidimensional scaling
  • Pattern recognition
  • Time series analysis and forecasting
  You may choose a non-statistical problem.
• Choose a package for R. Some examples are gbm, xgboost, LiblineaR, cluster, MASS, smacof, superMDS, neuralnet, deepnet, Rssa.
• Find a dataset for demonstrating how to use the chosen package for solving the chosen problem. Usually, each package contains references to few suitable datasets.
• Your group should write the choice of a problem, a package and a dataset in "Discussion" at Learning Central in order to avoid the same choice by other groups. On selection of a topic, it is advisable to ask Andrey Pepelyshev whether or not it is suitable.
• In your group coursework, (i) explain a problem, (ii) explain certain technical aspects of a package, (iii) explain solution of a particular problem for a dataset.

You are not constrained by the use of slides (although you are welcome to). Feel free to be imaginative.