Podchorążych 1, Sala 101, Kraków
We are happy to announce eleventh eRka meetup. This time we will have opportunity to listen about air pollution (PM10) in a more quantitative way than usually presented in mass media. The second presentation will be about C++ inside R.
If there will be interest from non-polish speakers then presentations will be delivered in English.
18:20 – 18:30 – Welcome
18:30 – 19:10 – “PM10 concentrations in Kraków - climatology of the air pollution and possibilities of short-term forecasting” – Bartosz Czernecki
19:10 – 19:40 – Pizza
19:40 – 20:10 – "I feel need. The need for speed. In other words - C++ inside R" – Zygmunt Zawadzki
BIOs and presentations abstracts
Atmospheric researcher, an expert on processing, programming and visualization of meteorological data. Currently working at the Department of Climatology at Adam Mickiewicz University in Poznań. Using R as primary data science tool since 2009. His scientific interests concerns mainly on modeling of atmospheric processes with use of empirical-statistical and dynamical downscaling approaches. He is also active in the topics of: extreme weather phenomena, renewable energy, air pollution and climate variability.
Presentation: PM10 concentrations in Kraków - climatology of the air pollution and possibilities of short-term forecasting
Central European agglomerations need to deal with the problem of a large concentration of particulate matter less than 10 or 2.5 micrometers in diameter (PM10 and PM2.5). This problem is of particular importance in Kraków which occupies top positions in the ranking of most polluted air in terms of PM10 concentrations (EEA 2015). In this presentation some regularities of PM10 concentration and its seasonality (sub-daily, weekly and in annual scale) are shown and compared with other Polish cities. Correlation analysis is conducted to show impact of meteorological elements causing high PM10 concentrations in Kraków, and to underline the role of human vs. environmental conditions causing exceedance of the PM10 legal concentration threshold and termination of long-lasting high-concentration episodes.
Some introductory results on possibilities of short-term forecasting of PM10 concentration with the use of hybrid numerical weather prediction system and R machine learning algorithms are going to be shown. The problem of data quality, data pre-processing and the uncertainty related with specifity of this issue are briefly discussed together with problem of visualization techniques required for different groups of end users.
R programmer with some experience in C++. Statistical gun for hire see - https://www.linkedin.com/in/zygmunt-zawadzki-81525b92
I feel need. The need for speed. In other words - C++ inside R
R's interpreter takes care about all dangerous stuff like memory allocation, therefore in R it is quite hard to shoot yourself in a foot. 'Stack overflow', 'segmentation fault' or 'memory leak' - such expression are not (happily) in the R's dictionary. But there's no free lunch. Protection costs CPU cycles. And sometimes you just need pure speed without any protection. In such situations C++ is a perfect choice...
In C++ shooting yourself in a foot means the loss of the whole leg. Programmer is closer to bare metal - it's more dangerous place, without any interpreters, and other stuff wasting precious CPU time to protect you (I exaggerate a bit:)).
This presentation will be about connecting R and C++ to achieve powerful workhorse for solving near any problem with secure and efficient way. I will try to convince R's users that learning C++ is a really good investment!