Return to STA410 Programming Portfolio Parent Repository here.
Submit sta410hw3.ipynb
on MarkUs before the end of the calendar day (EoD) on the due date.
Programming Portfolio Assignment 3 addresses optimization with a particular focus on Newton's Method and TensorFlow for Gradient and Hessian (Jacobian) computations.
- Root-finding with Newton-Raphson and fixed-point iteration and acceleration
- Iteratively reweighted least squares for fitting logistic regression models
- Nonlinear Gauss-Seidel using TensorFlow for partial derivative calculations
- Newton's method for arbitrary dimension gradients/Hessians with TensorFlow
UofT students may access this the collection of programming problems with the UofT JupyterHub or UofT JupyterLab via
[JupyterHub] https://jupyter.utoronto.ca/hub/user-redirect/git-pull?repo=https://github.com/pointOfive/sta410hw3&branch=main
Some notes to faciltate getting started in this environment are available on the UofT JupyterHub support page. If for some reason JupyterHub/Lab is not loading the repository, you can delete and reload repositories (after downloading and saving your work).
From JupyterHub/Lab, open a new terminal with
New
>Terminal
and then useyes y | rm -r <path to directory to delete>
to a delete the repository directory.
Alternatively, the programming problems may also be accessed without UofT authentication using Google Colab via
https://colab.research.google.com/github/pointOfive/sta410hw3/blob/main/sta410hw3.ipynb
If you're working in some other environment, the versioning there must support notebook format 4.5, e.g., JupyterLab >= 3.0.13 (for "JupyterLab UI") or Jupyter Notebook >= 6.2 (for "Jupyter classic UI"); otherwise, your notebook cell-ids will not be supported and you will not get any credit for your submitted work.
You may check if cell ids are present or changing at each save with
! grep '"id":' <path/to/notebook>.ipynb