| WEB | http://users.jyu.fi/~jutakarv/ |
In longitudinal studies, dropout leads to a monotone missing data pattern. We show in a new article that monotonicity sometimes enables and sometimes prevents the identification of the full law, i.e., the joint distribution of actual variables and response indicators.
We analyze the relationship between the value of information (VOI) and the decision-maker’s risk aversion in the lake monitoring where one needs to decide about whether to implement costly management actions or not. We calculate the value of perfect as well as imperfect information for risk neutral and risk averse decision-makers. The risk aversion is measured using the certain equivalent and the Arrow-Pratt risk aversion measures, which are defined using the derivatives of the utility function. We consider two utility functions for a risk averse decision-maker, an exponential utility and a power utility function, and demonstrate their use with lake management data from Finland. The results show that, in this context, a risk averse decision-maker’s VOI may be lower or higher than a risk neutral decision-maker’s VOI, depending on the prior probability of the lake being in the need of management actions and the cost of the actions. The risk aversion seems to have a clear impact on decisions. This may encourage the decision-makers to contemplate their risk preferences instead of hastily assuming the risk neutrality.
ELLIS Institute Finland is hiring Principal Investigators in AI + machine learning. World-class resources for research incl. LUMI supercomputer, generous starting package & professorship affiliation with a university in the world’s happiest country!
Apply by March 9: https://www.ellisinstitute.fi/PI-recruit
Adaptive Markov chain Monte Carlo (MCMC) algorithms, which automatically tune their parameters based on past samples, have proved extremely useful in practice. The self-tuning mechanism makes them `non-Markovian', which means that their validity cannot be ensured by standard Markov chains theory. Several different techniques have been suggested to analyse their theoretical properties, many of which are technically involved. The technical nature of the theory may make the methods unnecessarily unappealing. We discuss one technique -- based on a martingale decomposition -- with uniformly ergodic Markov transitions. We provide an accessible and self-contained treatment in this setting, and give detailed proofs of the results discussed in the paper, which only require basic understanding of martingale theory and general state space Markov chain concepts. We illustrate how our conditions can accomodate different types of adaptation schemes, and can give useful insight to the requirements which ensure their validity.
Simulating values from a known distribution is a basic task in statistics. But how to simulate from a counterfactual distribution? We consider this question in a recently published paper.
https://jair.org/index.php/jair/article/view/15579/27053
The proposed algorithm is applied to fairness analysis in credit-scoring.
Applications are open for the doctoral program in AI!
🎓 Fully funded PhD positions across 10 Finnish universities
🧭 Collaboration with industry
📹 Professor Arno Solin introduces the program in this video 👇
Apply here by Sept. 9 ➡️ https://fcai.fi/doctoral-program
#phdPosition #doctoralprogram #phd #artificialintelligence #machinelearning #finland #studyinfinland #research
Looking for a fully-funded PhD position in AI or ML? We are opening the next call for applications in the Finnish AI doctoral program soon!
See all the details and how to apply: https://fcai.fi/doctoral-program
#phdPosition #phd #academicjobs #artificialintelligence #machinelearning #finland #recruitment
dynamite: Bayesian Modeling and Causal Inference for Multivariate Longitudinal Data (built with Stan) by Santtu Tikka and @jouni_helske
... complex panel (time series) data comprising of multiple measurements per multiple individuals measured in time via dynamic multivariate panel models...
Easy-to-use and efficient interface for Bayesian inference of complex panel (time series) data using dynamic multivariate panel models by Helske and Tikka (2022) <doi:10.31235/osf.io/mdwu5>. The package supports joint modeling of multiple measurements per individual, time-varying and time-invariant effects, and a wide range of discrete and continuous distributions. Estimation of these dynamic multivariate panel models is carried out via Stan. For an in-depth tutorial of the package, see (Tikka and Helske, 2023) <arxiv:2302.01607>.
ELLIS Institute Finland
Matti Vihola
Finnish Center for AI
Stan