This is an interesting question asked by a director managing a team of scientists, who wants to keep his team updated on data science techniques. Below is the email exchange in question:
Hey Vincent. I've been doing presentations for my company for the last 3 months and I'm running out of topics. I hit regression, clustering, neural nets in general, boa, global search algorithms. Most of them have strong stats or data analytics backgrounds... any suggestions for a clinical pharma company group? I've got a lot of PhD psych and stats guys and I'm trying to broaden their horizons a bit.
What about survival models, decision trees, SVM, factorial analysis, time series (Box & Jenkins), analysis of variance / covariance, kriging (spatial stats), MCMC (hierarchical Bayesian models), graph theory, combinatorics, structural equation modeling, ensemble models, stochastic processes (Markov, birth and death, Brownian motion), ROC curve, discriminant analysis, density estimation, Monte Carlo simulations (anything related to operations research), experimental design (DOE), general linear model, are just a few that pop to my mind.
Anything else that you would recommend?
I would like to add one simple/traditionnal approach that should help : It comes from "french statistics school". Factoriel Analysis like PCA and MCA.