Skip to content

This is the accompanying GitHub repository to the Fall 2024 section of AIPI 510 in the Duke Masters of Engineering in Artificial Intelligence

Notifications You must be signed in to change notification settings

ludaladila/aipi510-fall24

 
 

Repository files navigation

Team Assignment #5

Statistical Analysis

Instructions

  1. Put together a code demo for your assigned topic.

Code should be:

  • Clean and well organized script
  • Using best practices (if you aren’t sure, go back to the Premodule content)
  • Well-commented
  • Contains appropriate unit testing
  • Clear name (ie ‘wilcoxon-test.py’)
  1. In addition to your code demo, you should put together a creative demonstration of your topic. Options include: song, rap, poem, children’s book, painting, short movie, interactive webpage, app demonstrating concept. You are welcome to do something beyond this list, just clear it with me first. You can use GenAI to create images for your creative component. If you use GenAI for your text component, you will need to go the extra mile and do a performance with it (ie a slam poem or sing the song and play guitar). Make sure to cite the GenAI used, per the syllabus.

Submission

To submit your code, make a PR into the statistical-analysis-ta5 branch and add me and the TA as reviewers. In your PR, add any links to your creative component. Also, add any requirements (and versions) that are not currently in the requirements.txt file to the text of your PR.

You will be presenting your creative component in class live either Week 6 or Week 7. If the creative component is a short movie, it is acceptable to show the video.

Topics

Descriptive Statistics

  • Measures of central tendency (mean, median, mode)
  • Measures of dispersion (standard deviation, variance, range, interquartile range)
  • Measures of distributions (skew, kurtosis)

Hypothesis Testing

  • Type I and Type II errors
  • P-values and significance levels
  • Multiple hypothesis correction (Bonferroni)

Parametric Tests

  • Z-score and standard normal distribution
  • One-sample t-test
  • Independent samples t-test
  • Paired samples t-test

Nonparametric Tests

  • Mann-Whitney U test (unpaired t-test)
  • Wilcoxon signed rank test (paired t-test)
  • Chi-square test for independence

ANOVA

  • One-way ANOVA
  • Post-hoc tests (Tukey)

Regression Analysis

  • Simple linear regression
  • Ordinary Least Squares (OLS) method and assumptions
  • Interpretation of regression coefficients
  • Multiple linear regression

Bayes’ Theorem

Model Evaluation

  • Goodness of fit measures (R-squared, AIC, BIC)
  • Residual analysis
  • Confidence Intervals
  • Simpson’s Paradox

Rubric

Code (15 points)

  • Code is a script, not a notebook
  • Code is clean and well organized
  • Code is documented with docstrings and comments
  • Code is free of commented out code (ie debug print statements)
  • Script has a clear name
  • Branching and PRs were done appropriately
  • Requirements are included in the text of the PR and are correct and versioned
  • The code runs as documented

Creative Component (30 points)

  • Creative Component is presented in class on the correct date (either Week 6 or 7, depending on topic)
  • Creative Component demonstrates topic in a clear way
  • Creative Component is creative

About

This is the accompanying GitHub repository to the Fall 2024 section of AIPI 510 in the Duke Masters of Engineering in Artificial Intelligence

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 51.2%
  • Jupyter Notebook 48.8%