Statistics and data science provide keys for understanding our increasingly complex world
The amount of information – and misinformation – in our world is growing at great speed. Collecting data is also easier than ever before.
Methods for analysing large masses of data are constantly being developed. By producing more up-to-date knowledge, we can more accurately estimate the probabilities of events, leading to more concrete impacts on society through, for example, policymaking.
Causal inference helps predict phenomena and events
Causality is a core concept in science. Causal relationships depict cause–effect relationships between different variables and help make inferences on how a specific factor (cause) impacts another factor (effect).
The inference of causal relationships, however, is not always straightforward.
A classic example of spurious correlation can be found by combining weekly data on ice-cream consumption with the number of drowning accidents.
In this example, weekly data on ice-cream consumption is contrasted with the number of drowning accidents. The data shows that both of these have peaks in summertime, but this does not mean, however, that there is a causal link between them. An explanatory factor underlying both these phenomena is the warm summer weather.
When examining human populations, for example, age, education and income level are factors that can explain the connections between many things.
Statistical thinking and inference are needed in almost all fields of science
In Finland, the Ģֱ is a leader in the field of causal inference research in statistics. Researchers from the Department of Mathematics and Statistics investigate in what kind of settings some interesting causal effect can be determined by means of available data sets.
The methods are applicable in many fields of science, says Senior Lecturer of Statistics Santtu Tikka from the Ģֱ.
The new type of methods we have developed make it possible to combine experimental and observational data sets and take into account missing data and selection effects as well."
The methods can be used in almost all fields of science, including environmental sciences, ecology, health sciences, social sciences, psychology, and bioinformatics.”
As a researcher of statistics, Santtu Tikka seeks to make causal research as simple and practical as possible. He has studied, for instance, identification algorithms for causal effects and how these could be improved in terms of both usability and statistical estimation.
Statistician can cooperate with experts from a range of fields
Methods arising from Tikka’s research are available as open-source software.
“I am fascinated by the endless possibilities of statistics,” Tikka says. “Many phenomena, such as climate change, are such that they cannot be studied experimentally. Then we must rely on observational data, but by means of appropriate methods causal inferencing is sometimes possible in these settings as well.”
Tikka has been involved in a joint project examining the impacts of legal reforms on the life of families, for example. The project has investigated, among other things, the causes and effects of the father’s use of family leaves. The project developed new statistical models applicable to causal inference, in particular.
In Tikka’s opinion, the best parts of his work are explorations with data and working with different people.
“It’s really enlightening and interesting to consider statistical problems together with experts from different fields,” Tikka says.
“Once we eventually find a common language and the statistical problem of the research is clarified, we can progress with the data, and by means of statistics we can finally find an answer to the research question.”
Tikka has already received a number of awards for his research. The Finnish Statistical Society granted him the Leo Törnqvist Award in 2015–2016 for the best master’s thesis in the field of statistics. The same society also granted him a doctoral dissertation award for 2017–2020.
Doctoral researcher in statistics feels he is making the world better
At the time of his master’s degree studies, Otto Tabell, who is now working on his doctoral dissertation at the University ofJyväskylä, had a summer job at the pharmaceutical company Orion, where he examined the effect of a specific medicine’s dosing on mortality rates.
The data concerned led eventually to his master’s thesis.
“I analysed the data by means of causal inference and created a causal model for the impact of medicine doses on the 90-day mortality rate,” says Tabell.
“By means of that model we could respond to certain problematic points observed in the analysis of previous studies.”.
Now he is preparing his doctoral dissertation on a causality theme in the doctoral education pilot entitled “Finnish Doctoral Program Network in Artificial Intelligence (AI-DOC)”. One of his research questions is how several different data sources can and should be combined to determine causal effects with clustered variables.
“If we wish to find out, for example, what impact added salt has on blood pressure, the causal effect can be identified by combining experimental research data and register data, for example,” Tabell says.
“Similarly, in causal inference the clustering of variables means that a group of several variables is examined as one variable. In the above-mentioned salt example, this kind of a variable cluster could consist, for instance, of various background variables collected from the research subjects.”
Statistical research calls for, above all, long-term orientation. It is often necessary to explore different approaches to elicit the results.
It is also great to get a chance to do meaningful work,” Tabell says.
“For example, by modelling consumers’ purchase decisions or figuring out background factors for environmental changes, it is possible to come up with new kinds of recommendations and make the world better.”
JYU to offer Master of Technology degrees in Statistics and Data Science
At present, statistics and data science can be studied at JYU, leading to the respective degrees of Bachelor of Science and Master of Science.
In 2026 the Ģֱ will launch a new Master of Technology programme in Statistics and Data Science.
The new Master of Technology programme responds to the shortage of skilled professionals, and it draws on the University’s strengths in statistics.
“At JYU, statistics and data science education is based on the department’s strengths as well as on cooperation with the Faculty of Information Technology,” says Professor of Statistics Juha Karvanen, from the Ģֱ.
“The Master of Technology programme will highlight probability-based inference, AI and the analytics of large data masses. The programme is more sensitive to the competence needs of companies.”
The field offers excellent career prospects
In the field of statistics and data science, employment is guaranteed, as the amount of data is growing all the time, and processing it requires statisticians. Students easily find summer jobs in their own field, which then often lead to an actual job after graduation. Cooperation with different companies and organisations is part of everyday activities in this field.
We provide plenty of master’s theses and research cooperation with various companies and organisations,” Karvanen says.
“The students get a foothold into the working life already during their studies.”
The Department of Mathematics and Statistics also supports their students’ practical training by granting them subsidies for it.
