GEOG5404M Analytics for Urban Policy

Urban Policy Briefing and Data Science Notebook

Coursework Brief

This assessment asks you to develop an in-depth data analysis into an urban policy area. The specific route that this analysis takes is for you to choose – so you will decide on the policy area and find and analyse the data. Inherent to this assessment is a challenge to your ability to identify an important policy area and an appropriate dataset and deliver a robust and useful analysis. Your analysis might seek to assess the impact of a previous policy or predict the implementation of a new one.

We anticipate that your learning from the formative assessments, in essential planning analyses like these, will be useful in shaping the design of this analysis. And feel free to implement one of those proposals for this assessment. We also encourage you to incorporate aspects from the field work within your analysis planning – and we would particularly be keen to see analyses that ask novel questions with urban analytics.

For the submission, you will be required to include two components, which will be assessed together. Different components of the report and notebook will attract differing weights, you should consult the marking rubric for full details.

Policy Briefing – a 2000-word written description that includes a) description of the urban policy context or challenge, b) details of prior work and initiatives around this theme (i.e., a literature review), c) a brief description of your analysis plan (this will be elaborated in the notebook), and) a discussion of the findings in the context of informing urban policy, including the inherent limitations and biases. You should refer to elements of the code notebook from your report to justify your discussion. The policy briefing should be prepared in Word (or similar) software and submitted via Minerva.

Code Notebook – a clearly structured Jupyter Notebook that shows the complete data analysis workflow. This is expected to include library loading, stages of data exploration, data wrangling and cleaning, and the data analysis process (e.g., results, visualisations, parameterisation, tests, etc.). You should accompany the methodology and analysis results with text, which aid navigation and reading of the process. The full interpretation of the results as a whole should be kept within the report section. Important – see note below about requirements for your notebook submission.

We are expecting to see an analysis plan that incorporates a sophisticated level of data science. Therefore, if you were to consider a retrospective analysis of policy, you might consider clustering or linear regression models as an endpoint to your workflow, whereas prospective analysis could involve implementation of non-linear regression or classification models. Data science workflows will be expected to be constructed carefully, demonstrating awareness of stages including (where appropriate) model and feature selection, normalisation, dimensionality reduction, hyperparameterisation, calibration, validation, and visualisation.

In total, the assessment will contribute 70% of your final grade for the module. Both components of the assessment should be submitted via Minerva.

The due date for this assignment is Friday 21th May 2024 at 2pm.

Further Notes and Guidance

Notebook submission requirements

Within a zip file, please include the following components:

· Your Jupyter notebook containing the complete analysis.

· Your data – or if the original dataset is too large, a processed version of it may be submitted, as long as the stage at which the data is exported from your workflow clearly marked.

· A YML file for your Conda environment, detailing the libraries and version numbers used for this analysis. You can export it by running this command from CL/Terminal (replacing ENVNAME as appropriate): conda env export --name ENVNAME > envname.yml

It is your responsibility to ensure that the notebook can run on another machine. Therefore, you should ensure all data are provided with relative directories accessible from the notebook.

Feedback

Feedback on your coursework will be returned to you 15 working days after the submission date. In cases where this is not possible, we will inform. you.

Word limit

The word limit for the written report is 2000 words. This limit includes EVERYTHING apart from i) the overall title of the piece of work; ii) where appropriate, an opening contents page; iii) the reference list or bibliography, iv) tables and figures (although captions for figures and tables are part of the word count), and v) any appendices. Tables should not contain lengthy passages of text in an attempt to circumvent the word limit; such cases will be investigated for academic malpractice. Please note that there is no leeway for word length. Marks will be deducted for work which is over the limit as set out in the module handbook.

Individual work and academic integrity

This is an individual assignment. Whilst it is natural to consult with your neighbour in the classroom, or to help each other with unfamiliar software/laboratory/field work, itis not permitted to share research, ideas, data or text that form. any part of this report/essay.

Students must also be careful to avoid plagiarism. Plagiarism means presenting the ideas of others as if they are your own. Examples of plagiarism include failing to cite the sources of ideas and information that are not your own and copying sections of text from sources that you have read. Collusion and plagiarism are both serious type of academic malpractice. If we suspect that these or other breaches of academic integrity have occurred, then they will be investigated according to university protocols.

For LUU information and advice on ‘Plagiarism, Fraudulent or Fabricated Coursework and Malpractice’ see:

https://leedsuniversityunion.knowledgeowl.com/help/plagiarism

For Skills@Library information on academic integrity and how to avoid plagiarism see:

https://library.leeds.ac.uk/info/1401/academic_skills/46/academic_integrity_and_plagiarism

Example Notebooks

Listed below are a number of example data analysis projects using Python and various libraries, combining code and narrative (to varying extents) within notebook format. In general, we expect a more systematic and complete analysis than that offered here – following the steps outlines above.

· Analysing police activity logs –

https://github.com/FraManl/DataCamp/blob/main/DS%20Project%20-%20Analyzing%20Police%20Activity.ipynb

· Brooklyn Bridge Pedestrian Traffic Analysis –

https://github.com/yagnesh7/doing_data_stuff/tree/master/brooklyn_bridge

· Buzzfeed analysis of Segregation in St Louis –

https://github.com/BuzzFeedNews/2014-08-st-louis-county-segregation/blob/master/notebooks/segregation-analysis.ipynb - needs better visualisation and documentation!

· Logistic models of well switching in Bangladesh –

https://github.com/pyro-ppl/numpyro/blob/master/notebooks/source/bayesian_hierarchical_stacking.ipynb - lacks descriptions of the data

· Clustering Samsung smartphone accelerometer data –

https://github.com/microsoft/Reactors/blob/main/workshop-resources/data-science-and-machine-learning/Data_Science_2/human-behavior-project/human_behavior_activity.ipynb

Further Details

Please refer to the module handbook for full details, including referencing, submission, late penalties, and word count policy.