2 resultados para Threshold regression

em Universidade Federal do Pará


Relevância:

20.00% 20.00%

Publicador:

Resumo:

Este trabalho faz uma análise do processo de convergência de renda entre os municípios da Amazônia Legal no período de 1980 a 2004, no intuito de identificar a existência de clubes de convergência com base em um modelo threshold. Os resultados indicam a formação de seis clubes de convergência, evidenciado-se um processo de convergência entre os clubes extremos. De uma forma geral, o capital humano mostra-se significante para explicar o processo de crescimento dos grupos de municípios mais ricos. Além disso, os resultados indicam que o capital físico foi significativamente negativo para explicar o processo de crescimento econômico da maioria dos clubes de convergência, o que demonstra o papel detrimentoso do investimento público realizado nos municípios da região.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

The purpose of this study was to compare contrast sensitivity estimated from transient visual evoked potentials (VEPs) elicited by achromatic pattern-reversal and pattern-onset/offset modes. The stimuli were 2-cpd, achromatic horizontal gratings presented either as a 1 Hz pattern reversal or a 300 ms onset/700 ms offset stimulus. Contrast thresholds were estimated by linear regression to amplitudes of VEP components vs. the logarithm of the stimulus contrasts, and these regressions were extrapolated to the zero amplitude level. Contrast sensitivity was defined as the inverse of contrast threshold. For pattern reversal, the relation between the P100 amplitude and log of the stimulus contrast was best described by two separate linear regressions. For the N135 component, a single straight line was sufficient. In the case of pattern onset/offset for both the C1 and C2 components, single straight lines described their amplitude vs. log contrast relations in the medium-to-low contrast range. Some saturation was observed for C2 components. The contrast sensitivity estimated from the low-contrast limb of the P100, from the N135, and from the C2 were all similar but higher than those obtained from the high-contrast limb of the P100 and C1 data, which were also similar to each other. With 2 cpd stimuli, a mechanism possibly driven by the M pathway appeared to contribute to the P100 component at medium-to-low contrasts and to the N135 and C2 components at all contrast levels, whereas another mechanism, possibly driven by the P and M pathways, appeared to contribute to the P100 component at high contrast and C1 component at all contrast levels.