Panagiotis (Panos) Toulis
Assistant Professor & John E. Jeuck Fellow
Econometrics and Statistics
- (2015) sgd R package at CRAN here.
(2013) Newer software on Github. I have moved all software projects in Github. Please check out this page https://github.com/ptoulis for code in causal inference, robust stochastic learning, kidney exchanges, or other side projects.
- (2010) Perl implementation of the LDA model (Latent Dirichlet Allocation).
I built A Perl implementation of the LDA topic model (Blei, Ng and Jordan (2003)). Download the package and follow the instructions on how to build and run LDA instances.
- Sentiment analysis for finance applications: I wrote software using Perl/MPI in order to
process in parallel multiple news sources and perform sentiment analysis
for two pre-defined assets (e.g. dollar vs. euro). The output is a histogram-like plot indicating
positive/negative views with green/red colors. Check out a demo video here.
The model used to "understand" the natural-language text is novel and based on "Meaning Filtering Trees" (MFT)
which aim to associate descriptive adjectives/verbs (e.g. "plummet", "fall", "rise", "in trouble" etc) with the appropriate
entities (e.g. "dollar", "euro" etc) and quantify their positive/negative sentiment.
- Evolutionary Dynamics - 'In Silico Demo':
A simple dynamic demo for viewing the evolution of reactive agent strategies (Tit-for-Tat, All-cooperative etc.)
Most of times you will notice the gradual behaviorial shift from defecting to co-operating and forgiving inside the population.
This is based on Martin Nowak's "Evolutionary Dynamics" book.
- Optimizing Scrip Systems
In short, scrip systems can be used in order to battle
The Tragedy of Commons.
The Birth-Death process involves an initial population and some stochastic process in which, each single element in the population either gives birth to further offsprings, or dies, or stays inactive.
Explore a Birth-Death process with only one element at the beginning. Use the "Draw" button, to plot the population over time.