This blog post presents my Master thesis (available here), where I created a new measure of interstate hostility by applying Bayesian ordinal item-response theory (O-IRT) model to a conflict events dataset, which I have created using data from Militarized Interstate Disputes (MID), Integrated Crisis Early Warning System (ICEWS), and International Crisis Behavior (ICB) datasets.

This post is the result of my homework for IST: 597 Foundations of Deep Learning class at Penn State. In the first part, I will show how I develop softmax regression model and multilayear perceptron (MLP) to solve the XOR problem. In the second part, I will fit a two hidden layer MLP to a version of the IRIS dataset.

This blog post is designed to give an introduction of how to estimate Item-Response Theory (IRT) model in R using the RStan package.