Data Analysis Project with Reports

Project requirements
1) Submission package

Report files

Part 1 report: maximum 2 pages

Part 2 report: maximum 6 pages

Page limits include figures and tables

Page limits exclude title page, table of contents, and references

The report must include explanations suitable for readers not familiar with programming

Report must be submitted as PDF (or other allowed document format as specified by your portal)

Code files (reproducible)

Python code must be provided as Jupyter notebook(s) (.ipynb)

R code must be provided as RMarkdown (.Rmd) if required by the brief

Code must be runnable by someone else to reproduce your results

If you reuse outputs or cleaned data from earlier questions, you must reference them clearly

Submission location

All relevant files must be submitted to the designated Canvas/VLE portal

2) Part 1 requirements (Random Walk Metropolis)

Part 1(a)

Target density: f(x) = (1/2) exp(-|x|) (Laplace(0,1))

Set an initial value x0

Use N = 10,000 iterations and proposal standard deviation s = 1

Use samples x1,…,xN (exclude x0)

Outputs for Part 1(a)

One plot containing:

histogram of samples

KDE of samples

true density curve overlay

Monte Carlo estimates:

sample mean

sample standard deviation

Implement the acceptance decision using the log-acceptance form for numerical stability

Part 1(b)

Run multiple chains (J chains), potentially from different initial values

Required case: N = 2,000, s = 0.001, J = 4

Compute:

per-chain mean Mj

per-chain variance Vj using denominator N (not N−1)

within-chain average W

between-chain variance B

Rb = sqrt((B + W) / W)

Outputs for Part 1(b)

Report Rb for the required case s = 0.001 and interpret it against the threshold

Sweep s from 0.001 to 1 and compute Rb(s)

Plot Rb against s

Identify the smallest s such that Rb < 1.05

From your R run: smallest s with Rb < 1.05 is 0.1701254

3) Part 2 requirements (Flights dataset analysis)

Core tasks

Use the ASA flights dataset (or a justified subset/time window)

Answer the Part 2 questions, including:

best times/days to minimise delays

relationship between plane age and delays

yearly logistic regression for diverted flights and coefficient trends across years

Reporting requirements for Part 2

Document all steps from raw data to final answers:

data loading and storage choices (files, database, etc.)

cleaning and wrangling operations

modelling decisions and justification

results presented with appropriate tables/figures

Ensure reproducibility with clear code organisation and comments

4) Consistency and organisation requirements

Do not mix results from different runs without saying so (figures and reported numbers should come from the same run)

Use clear figure numbering and captions (Figure 1, Figure 2, etc.)

Use consistent file naming so it is obvious which file corresponds to which part/question

If you re-upload the coursework PDF/brief, I can rewrite this as a strict checklist that matches the official wording exactly.