Statistically Insignificant

Languages, Statistics and other things

| Comments

In this post I will briefly introduce the EM algorithm with two simple examples. The EM algorithm uses an iterative approach to find the Maximum Likelihood estimate for a model with latent variables.

Note I will not provide a thorough coverage of the mathematics but rather focus on two examples of Gaussian Mixture Models.

| Comments

In this post I list some resources which have helped me to learn Spanish, I hope that they will help you too!

All of the listed resources are (mostly) free, except News In Slow Spanish (and the Latino version), and clearly the textbooks, TV series and films (of course!).

| Comments

Introduction

Mendelian Randomization is an approach to test for a causal effect from observational data in the presence of certain confounding factors. It uses the measured variation of genes (of known function) to bound the causal effect of a modifiable exposure (environment) on a phenotype (disease). The fundamental idea is that the genotypes are randomly assigned (due to recombination in meiosis under certain assumptions), and this allows them to be used as an instrumental variable.

Introducción

La aleatorización mendeliana es un método para confirmar un efecto causal de datos observacionales, posiblemente en la presencia de factores de confusión.

El método utiliza la discrepancia medida de genes (de funciones ya sabidas) para limitar la influencia causal de una exposición (por ej. el ambiente) en un fenotípo (por ej. una enfermedad). La idea fundamental es que los genotípos son asignados aleatoriamente (por la recombinación durante la meiosis), y esto permita que se puede ser utilizados como variables instrumentales.

| Comments

I’ve started this blog to save resources which have helped me in learning languages, concepts in statistics, and anything else which I think might help others too.

I am currently a PhD student in Tübingen, Germany in “Machine Learning for Personalised Medicine”. I’m particularly interested in models combining genotypic and phenotypic measurements (such as gene expression, or methylation), in addition to the application of causal inference methods to justify causal conclusions from observational data, where possible.

I also intend to post summaries of my presentations and work here, as part of the “outreach” encouraged in my PhD programme.

I hope it will be useful to some others too!