We compared maximum likelihood and the k-NN algorithm in the context of discriminant analysis with spherical data.
Discriminant analysis for spherical data, or directional data in general, has not been extensively studied, and most papers focus on one distribution, the von Mises-Fisher. In this work, we study more distributions, escaping the rotational symmetry bound of the aforementioned distribution and also include a non parametric classier, the k-NN algorithm.
A folded type model is developed for analyzing compositional data based that provides a new and flexible class of distributions for modeling data defined on the simplex sample space. Despite its rather
seemingly complex structure, employment of the EM algorithm guarantees efficient parameter estimation.
We measure performance on the basis of a publishing productivity index which allows to account for difference in research inputs among departments.
This letter proposes a simple test for the linearity of a time series. We compare the small and large samples properties of the suggested test via Monte Carlo techniques with well known time domain linearity tests. Our results suggest that the suggested test over performs the power of the other competitive tests in small samples.
In this paper we investigate the effects of temporal aggregation and systematic sampling using some well known linear and nonlinear Granger causality tests.
This short paper demonstrates that the use of temporally aggregated data may affect the power and the size of the well known the Ramsey's (1969) RESET test.
This paper examines the existence of a linear or nonlinear interaction between the Advance/Decline ratio index and the returns of the Athens General Index.
This short empirical paper examines the unemployment dynamics in Greece both in the long run and during the current crisis.
Various methods have been developed to improve mortality forecasts. The authors proposed a neuro-fuzzy model to forecast the mortality. The forecasting of mortality is curried out by an ANFIS model which uses a first order Sugeno-type FIS.
The paper presents a new technique in the field of unemployment modeling in order to forecast unemployment index. Techniques from the Artificial Neural Networks and from fuzzy logic have been combined to generate a neuro-fuzzy model.
In this paper, we examine the effects of data collection frequency on the computation of the Consumer Price Index (CPI).
This short paper demonstrates the effects of using missing data on the power of the well-known Hausman (1978) test for simultaneity in structural econometric models.
The aim of the paper is to determine the state of IT within the Romanian organizations and its impact for the Romanian economy.
This short paper examines the nonlinear interaction between mutual fund flows and stock returns in Greece. We investigate the possibility of a nonlinear causality mechanism through which mutual funds flows may affect stock returns and vice versa.
A crucial aspect of empirical research based on ARIMA(p,q) model is the choice of the appropriate lag order. Several criteria have been used in order to identify the appropriate order of a ARIMA(p,q) process. In this paper we investigate the effects of using a variation of selection criteria under different temporal aggregation levels.
This letter proposes a simple test for the linearity of a time series. We compare the small and large samples properties of the suggested test via Monte Carlo techniques with well known time domain linearity tests.
In this short paper a Gamma distributed lags model is used to study the diachronic responses between the actual data and the forecasts supplied by OECD the last 27 years for the case of the Greek Economy.
This paper is using simple nonlinearity tests to provide evidence of a positive and significant causal relationship going from stock market development to economic growth in Greece during the last 10 years.
Join the notification list of the Department of Economics.