This document sets out the three (3) questions you are to complete for CAB420 Assignment 1A. Problem 1 highlights the content from weeks 1 and 2, Problem 2 from week 3, and Problem 3 from weeks 4 and 5; though material from all weeks will aid in addressing these problems. This assignment has been designed such that it can be completed solely using the resources covered in class, though students are encouraged to refer to external sources and conduct further research into the tasks and cite relevant sources.

The assignment is worth 15% of the overall subject grade. All questions are weighted equally. Students are to work individually.  Students should submit their answers in a single docu- ment (either a PDF or word document), and upload this through the Canvas submission link. Further Instructions:

1. Data required for this assessment is available on canvas alongside this document in CAB420 Assessment 1A Data.zip. Please refer to individual questions regarding which data to use for which question.

2. Juptyer Notebook python templates have been provided for each question on Cavnas.

3. Answers should be submitted via Cavnas.  In the event that Cavnas is down, or you are unable to submit via Cavnas, please email your responses to cab420query@qut . edu.au.

4. For each question, a concise written response.  Guidelines for the length of the written responses have been supplied in this assignment brief. These guides pertain to the written component of each section (ie. excluding figures), though figures should be included to enhance the response only.  Figures without sufficient discussion show no evidence of understanding of the content or the task. The length of responses may exceed these guides, but should do so sparingly and only when believed to be suitable. Note that the CRA requires responses to be “clear and concise”, so verbose and lengthy submissions that needlessly disregard these guides will be penalised according to the CRA. This similarly applies to the excessive use of figures without sufficient discussion or evaluation.

5. Responses should explain and justify the approach taken to address the question (in- cluding, if relevant, why the approach was selected over other possible methods), and include results, relevant figures, and analysis. Python Notebooks, or similar ma- terials will not on their own constitute a valid response to a question and will score a mark of 0.

6. Responses should highlight where the modelling approach has been successful, and to highlight the limitations apparent.  When limitations are encountered, students are encouraged to investigate and identify what the underlying cause of any failure or limitations are.

7. Python code, including live scripts or notebooks (or equivalent materials for other languages) may optionally be included as appendices. Figures and outputs/results that are critical to question answers should be included in the main question response, and not appear only in an appendix.

8. Students are encouraged to use any code presented in lectures/tutorials/examples from CAB420. Use of external code is also permitted, though requires to be correctly  cited and referenced.

9. Students who require an extension should lodge their extension application with HiQ (see http://external-apps .qut .edu .au/studentservices/concession/).   Please note that teaching staff (including the unit coordinator) cannot grant extensions.

Problem  1.  Regression. The data in the Q1 directory in CAB420 Assessment 1A Data .zip contains socio-economic data from the 1990 US census for various US communities, and the number of violent crimes per capita (in the column ViolentCrimesPerPop).  The purpose of the data is to explore the link between the various socio-economic factors and crime.

The provided data has been split into training (communities train),

validation (communities val), and testing (communities test) sets.

Your Task:   Given the provided data, you are to:

❼ Train a linear regression model to predict the number of violent crimes per captia from

the socio-economic data.

❼ Train a Ridge regression model to predict the number of violent crimes per captia from

the socio-economic data.

❼ Train a LASSO regression model to predict the number of violent crimes per captia

from the socio-economic data.

For your analysis, you are to use all provided data (i.e. DO NOT perform any filtering or selection to remove columns and/or rows).  For LASSO and Ridge models, the validation dataset should be used to select the optimal value of λ .  All models should be evaluated on the test set.  Evaluations should consider the predictive power of the model, the model complexity, and the model validity.

Your response must include sections that address the following:

❼ Discussion and justification of any pre-processing performed, such as standardisation.

(1/3 page)

❼ Details of the three trained models, including key parameters such as values for λ for

the LASSO and Ridge models, and a brief discussion of how these were selected. (1/3 page)

❼ An evaluation comparing the three models.  Your evaluation should consider model

accuracy and model validity. Discussion of accuracy/validity should also consider the socio-economic nature of this data, and if or how this impacts the the required accuracy and/or utility of the resultant models. You may also wish to discuss the relevance of terms within the model. The use of tables to summarise results, and figures to highlight details such as model validity, is strongly recommended. (1 page)

❼ Socio-economic modelling is incredibly complex and often dealing with sensitive data

and variables. Briefly discuss the ethical concerns that need to be considered when evaluating the models developed for this problem.  This discussion may include ref- erence to potential confounding variables, or to limitations in the modelling schemes used. Students are encouraged to highlight these ethical considerations and limitations to help avoid misinterpretation of the results. (1/3 page)

Problem  2.  Classification.  Land use classification is an important task to understand our changing environment. One approach to this involves the use of data from aerial sensors that captures different spectral reflectance properties of the ground below. From this data, the land type can be classified.

You have been provided with training,  validation  and testing data  (Q2/training .csv, Q2/validation .csv and Q2/testing .csv) that include 27 spectral properties and an overall classification of land type, which can be one of:

❼ s: ‘Sugi’forest;

❼ h: ‘Hinoki’forest;

❼ d: ‘Mixed deciduous’ forest;

❼ o: ‘Other’non-forest land.

Your Task:   Using the provided data as-is, you are to train three multi-class classifiers to classify land type from the spectral data. These classifiers are to be:

1. A K-Nearest Neighbours Classifier;

2. A Random Forest; and

3. An ensemble of Support Vector Machines.

Model hyper-parameters should be selected using a grid search operating over the validation set. The resultant models are to be evaluated on the testing set and compared.

Your response must include sections that address the following:

❼ Discussion and justification of any pre-processing performed, such as standardisation.

(1/3 page)

❼ Details of the hyper-parameter selection method, the final model parameters selected,

and a discussion of these in relation to characteristics of the data.  Discuss the effect of your hyper-parameters will have on the model (ie. if you have found a small value for a hyper parameter, what effect will this have on the model compared to a large value?) (2/3 page)

❼ An evaluation and comparison of the final three models, including a discussion explor-

ing any difference in performance between the models. (1 page)