computing

EM Benchmarks

I am getting into estimating mixture models at the moment. In particular in the context of models of wage formation where unobserved heterogeneity stemming from both firm and worker side is often modeled with a mixture model. The main assumptions are that Firms are classifiable into types $l \in \{1,\dots,L\}$, workers into $k \in \{1,\dots,K\}$ If Worker $i$ is of type $k$ and works for firm $l$ in a certain period, their wages are drawn from distribution $\mathcal{N}(\mu_{k,l},\sigma_{k,l})$.

Computing

You can find all of my publicly assessible software on my github profile. I have written software in high-level languages like Matlab, python, R and julia. I am an expert R and julia user, I’ve good python skills, Matlab is somewhat outdated. I am well-versed with C++ and fortran. I am using PostgreSQL + postgis databases (interfaced from either R or python) to manage large scale geo-spatial datasets. I have used the phantastic Rcpp (R + C++) package in several projects.