transfer-performance-project

This is the code repository for "Transfer Performance Evaluation Using Bayesian Methods."

The code can be run as follows.

All univariate results are entirely contained within univariate_normal/univariate_normal_linear.ipynb.

Similarly, all multivariate results are contained within running multivariate_normal/multivariate_normal_linear.ipynb.

The code is all in Python, and requires the following packages to run:

• numpy
• matplotlib (specifically, the pyplot module)
• random
• math
• pandas
• sklearn
• itertools
• scipy (specifically, the stats module)
• time
• statsmodels (specifically, the api module)
• cvxopt
• datetime (specifically the datetime module)

Since the random seed is fixed, so long as everything is ran once, results should be replicated. The end of univariate_normal/univariate_normal_linear.ipynb also contains the "easter egg" of an unfinished Poisson example that proved uninsightful.