Sam Andersson
I'm a |
About
Get to know me
Probabilistic models for partially observed, time-structured data.
I'm a computational statistician and doctoral researcher at Karolinska Institutet. I build probabilistic and statistical machine-learning methods — latent-state and semi-Markov models, Bayesian inference, simulation, and uncertainty-aware prediction — for large, noisy, individual-level longitudinal data.
Research
Selected publications
- All
- Published
- Preprint
- In preparation
Insights into the temporal dynamics of identifying problem gambling on an online casino
A machine-learning study of 4.5 years of individual online-casino account data, examining how early and how reliably problem gambling can be flagged — using leakage-controlled, temporally held-out validation and 30/60/90-day truncation analysis.
Finite-Probe Total-Variation Certificates for Finite-Basis Drifting Models
Develops a held-out total-variation certificate for finite-basis drifting models by expressing finitely sampled numerator drift as vec(VX) = M c. The certificate propagates field-estimation error, operator-calibration error, representation residuals, and probe conditioning, and abstains when the observability margin is nonpositive. The analysis also identifies rank and symmetry degeneracies and proves large-bandwidth collapse toward mean matching.
Heavy-tail-aware representation learning and dynamic Bayesian state modelling for an operational proxy of problem-gambling risk
Learns heavy-tail-aware temporal representations and fits dynamic Bayesian (Gaussian HMM) state models to derive an operational, capacity-aware proxy for problem-gambling risk from routine online data — under sparse, missing-not-at-random labels.
Mixed-emission hidden semi-Markov modelling of online gambling telemetry during a deposit-cap policy period
A mixed-emission hidden semi-Markov model — Gaussian and negative-binomial emissions, dwell-age hazards, forward likelihood recursion — for partially observed multichannel gambling telemetry, with diagnostic-gated Bayesian inference and simulation recovery.
Predicting women with depressive symptoms postpartum with machine-learning methods
Applies machine-learning methods to longitudinal clinical and epidemiological data to predict which women go on to develop depressive symptoms postpartum.
Contact
Get in touch
I'm always glad to talk through research problems, methods, or a possible collaboration. Email is the surest way to reach me.
- Email [email protected]
- Location Stockholm, Sweden
- GitHub SamAndersson-C