Code and Data

Technical Appendix for NMSC

Network Meta-Analysis Model and Code 
Associated papersAppendix G. Network Meta-Analysis Model and Code

ACL 2014 (irony)

Data (& code) for our ACL 2014 paper 
Associated papersHumans Require Context to Infer Ironic Intent (so Computers Probably do, too) 

JAGS code and R data for multinomial network model

JAGS code for multinomial network model for the example using partial data (five outcomes, three treatments) 
Associated papersBayesian network meta-analysis for unordered categorical outcomes with incomplete data 

JAS code

Source code for our Joint, Additive, Sequential model of topics and speech acts. 
Associated papersA Generative Joint, Additive, Sequential Model of Topics and Speech Acts in Patient-Doctor Communication 

Online physician reviews

A corpus of (50,000+) online reviews of physicians across multiple aspects of care. 
Associated papersWhat Affects Patient (Dis)satisfaction? Analyzing Online Doctor Ratings with a Joint Topic-Sentiment Model 

Imbalanced dataset

This is a collection of 35 imbalanced binary classification datasets. These have been preprocessed into feature-vectors (and ultimately into SVM-light style sparse-formatted files). 
Associated papersImproving Class Probability Estimates for Imbalanced Data 

Corpus of encoded, imbalanced binary classification datasets

A set of 35 datasets used in our work on class probability estimates in imbalanced scenarios. 
Associated papersClass Probability Estimates are Unreliable for Imbalanced Data (and How to Fix Them) 

Code for all-subsets meta-analysis

Stata c-plug-in to perform all-subsets meta-analysis. 
Associated papersGOSH – a graphical display of study heterogeneity 

Infinite Jest (annotated)

The text of infinite jest annotated with sub-narratives. 
Associated papersMultiple narrative disentanglement: Unraveling infinite jest 

Multiple time points

Stata code for "Meta-analysis of outcomes reported at multiple time points: a multivariate approach" 
Associated papersMeta-analysis of effect sizes reported at multiple time points: A multivariate approach