Matti Pajari
Research Subjects
Matti Pajari has a total of 3 co-authors. Here are the top co-authors by number of shared publications:
- Lasse Makkonen
VTT Technical Research Centre of Finland, Espoo, Finland
2 shared publications Maria Tikanmäki
VTT Technical Research Centre of Finland
1 shared publication- Maria Tikanmäki
VTT Technical Research Centre of Finland Ltd Espoo Finland
1 shared publication
Latest Publications
- Open Access
- 10 December 2019
- Published by Taylor & Francis in Communications in Statistics - Theory and Methods
- Vol. 50 (14) , 3319-3337
Probabilistic evaluation of quantile estimators
The foundations of the criteria to assess the goodness of quantile estimators for continuous random variables are reviewed and the probabilistic justification for a novel bin-criterion is presented. It is shown that the bin-criterion is a more appropriate measure of goodness of a quantile estimator than those based on minimizing the bias of the quantiles or the parameters of the distribution.
- 10 December 2019
- Published by Taylor & Francis in Communications in Statistics - Theory and Methods
- Vol. 50 (14) , 3319-3337
Open AccessProbabilistic evaluation of quantile estimators
The foundations of the criteria to assess the goodness of quantile estimators for continuous random variables are reviewed and the probabilistic justification for a novel bin-criterion is presented. It is shown that the bin-criterion is a more appropriate measure of goodness of a quantile estimator than those based on minimizing the bias of the quantiles or the parameters of the distribution.
- 1 September 2013
- Published by Canadian Science Publishing in Canadian Journal of Civil Engineering
- Vol. 40 (9) , 927-929
Discussion on “Plotting positions for fitting distributions and extreme value analysis”
Plotting positions are used in the extreme value analysis for many engineering applications. The authors of the discussed paper concluded based on their simulations that distribution dependent plotting position formulae provide a better fit to the underlying cumulative distribution than the distribution free Weibull formula. We show here by Monte Carlo simulations following the theory of probability that the opposite is true, and outline that the criteria used in the comparisons made by the authors of discussed paper are inappropriate. Accordingly, the Weibull formula should be used as the unique plotting position.
- 1 September 2013
- Published by Canadian Science Publishing in Canadian Journal of Civil Engineering
- Vol. 40 (9) , 927-929
Discussion on “Plotting positions for fitting distributions and extreme value analysis”
Plotting positions are used in the extreme value analysis for many engineering applications. The authors of the discussed paper concluded based on their simulations that distribution dependent plotting position formulae provide a better fit to the underlying cumulative distribution than the distribution free Weibull formula. We show here by Monte Carlo simulations following the theory of probability that the opposite is true, and outline that the criteria used in the comparisons made by the authors of discussed paper are inappropriate. Accordingly, the Weibull formula should be used as the unique plotting position.