Bulletin of Taras Shevchenko National University of Kyiv. Astronomy, no. 59, p. 34-36 (2019)
Some problems in the statistical processing of astronomic data
Parnovsky S., Dr. Sci
Astronomical Observatory, Taras Shevchenko National University of Kyiv
The problems that could arise with a correct statistical processing of astronomical data and a search for linear statistical dependences are analyzed. It has been shown that the use of standard statistical software can yield incorrect results. The iterative procedure of processing based on multi-year experience is proposed. It covers all stages: creation of sampling, discarding of outliers, search for statistically significant regressors, excluding insignificant regressors, finding out multiparametric dependencies and their analysis, etc. This paper is a methodological one and is it describes some potential traps in the statistical processing of observational data. We forewarn against some actions that may lead to data selection or to incorrect conclusions about the influence of some factors on the quantity being studied. Some examples of problems associated with data selection, correlation between regressors, Malmquist bias and correction are given.
Statistical processing, astronomic data
Parnovsky, S.L., Parnovsky,A.S. 2013, Astrophysics and Space Science, 343, 747
Fisher, R.A. 1954
Seber, Dzh. 1980, M.: Myr, 456
Hudson, D.J. 1970, М.: Myr, 1964
Press, W.H., Teukolsky,S.A., Vetterling, W T., Flannery, B.P., 1986, Cambridge University Press
Parnovsky, S.L., Parnovsky, A.S., 2010, Journal of Physical Studies, 14, 3, 3901.
Parnovskyi, S., Izotova, I., Parnovsky, S., Izotova, I., 2018, Visnyk Kyivskoho natsionalnoho universytetu imeni Tarasa Shevchenka. Astronomiia, V57(1), 41