Podatkovni analitiki, osebe, ki znajo s kombinacijo računalniškega in statistično/matematičnega znanja izluščiti koristne informacije iz podatkov, so danes eden izmed najbolj zaželenih in iskanih kadrov. Glavni cilj tega predmeta je študenta opremiti z znanjem, ki je potrebno za reševanje večine statističnih nalog, ki jih srečamo pri vsakdanjem statističnem delu in empiričnih raziskavah, obenem pa študentu predstaviti tudi teoretično in algoritmično ozadje, ki nam omogoča statistično analizo.

Najprej se bomo naučili temeljev statistike, natančneje Bayesove statistike. Bayesova statistika je sodoben statistični pristop in je v veliko pogledih bolj intuitivna kot klasična statistika. Še posebej pa je primerna za tiste, ki imajo programersko ozadje in so vešči algoritmičnega razmišljanja. Spoznali se bomo s koncepti statističnega modeliranja ter klasičnimi in sodobnimi algoritmov za statistično računanje (Gibbs, Metropolis-Hastings, Hamiltonian Monte Carlo). V drugem delu predmeta bo poudarek na uporabi pridobljenega znanja pri reševanju praktičnih problemov, od preprostih, pa vse do aktualnih raziskovalnih problemov. Od predpriprave podatkov, preko modeliranja, do predstavitve rezultatov. Pogledali si bomo tipične naloge napovedovanja, gručenja in testiranja hiptez na različnih področjih, kot so napovedovanje izidov športnih tekem, modeliranje poti zračnih mas in onesnaženosti zraka, gručenje delov možganov glede na podobnosti v aktivnosti in mnoge druge... Pri delu si bomo pomagali s programskim jezikom R in sodobnim orodjem za Bayesovo statistiko Stan.

Večina končne ocene bo iz vaj in domačih nalog, manjši del pa iz pisnega izpita. Študenti, ki se bodo izkazali v prvem delu predmeta, bodo v drugem delu predmeta imeli možnost pisni izpit nadomestiti z mentoriranim delom na uprabi statistike pri reševanju aktualnega praktičnega problema.Data analysts, people who are able to, with a combination of computer and statistical/mathematical knowledge, extract useful information from data, are today one of the most desired and sought-after employees. The main objective of this course is to equip the student for solving most data analysis tasks we
encounter in every-day statistical practice and empirical research,
while at the same time introducing the theoretical background and
algorithms that make statistical analysis possible.

We will first learn the fundamentals of
statistics, in particular, Bayesian statistics. Bayesian statistics is a modern approach to solving
statistical problems and is in many ways more intuitive than classical
statistics.
It is also well suited to those who have a background in
computer programming and algorithmic thinking.
We will get familiar with the concepts of statistical modelling, prior and posterior probability, and classical and modern algorithms for statistical inference (Gibbs, Metropolis-Hastings, Hamiltonian Monte Carlo). In the second half of the course emphasis will be on applying what we have learned to solving practical problems, from simple ones up to current research problems. From data preprocessing. through modelling, to presenting results. We will look at the tasks of prediction, clustering, hypothesis testing in various different fields, such as predicting the outcome of sports matches, modelling airflow trajectories and air pollution, clustering brain nodes with similar activity patterns, and many more... We will be using the R programming language and the modern Bayesian inference tool Stan.

Most of the final grade will be from graded lab-work and homework and a smaller part from a written exam. Students that perform well in the first half will be offered the opportunity to work, with guidance, on a real-world applied statistics problem instead of the written exam.