Basic Statistical Estimation Outperforms Machine Learning in Monthly Prediction of Seasonal Climatic Parameters

Machine learning (ML) has been utilized to predict climatic parameters, and many successes have been reported in the literature. In this paper, we scrutinize the effectiveness of five widely used ML algorithms in the monthly prediction of seasonal climatic parameters using monthly image data. Specifically, we quantify the predictive performance of these algorithms applied to five climatic parameters using various combinations of features. We compare the predictive accuracy of the resulting trained ML models to that of basic statistical estimators that are computed directly from the training data. Our results show that ML never significantly outperforms the statistical baseline, and underperforms for most feature sets. Unlike previous similar studies, we provide error bars for the relative performance of different predictors based on jackknife estimates applied to differences in predictive error magnitudes. We also show that the practice of shuffling data sequences which was employed in some previous references leads to data leakage, resulting in over-estimated performance. Ultimately, the paper demonstrates the importance of using well-grounded statistical techniques when producing and analyzing the results of ML predictive models.

Falconiformes assemblages in a fragmented landscape of the Atlantic Forest in Southern Brazil

An ecological analysis focused on distribution of Falconiformes, comprising abundance and morphology data, is provided. Samples were collected in a fragmented landscape within the Atlantic Rainforest, southern Brazil, between August and November 2001. Four main types of habitats were pinpointed among the 30 different sampling sites, while eight external morphological traits were employed. Twenty-one Falconiformes species were detected and jackknife estimates for regional richness reached 24.8 ± 2.56 species (p<0.05). There were no differences between average number of diurnal birds of prey species in the different habitats under analysis (H-test, p>0.05). Mantel's test for relative abundance and species morphology reveals weak association rates, corroborating the lack of association between matrixes (r = 0.059, p>0.05). Difficulties in the analyses of distribution and species morphology data may have been caused by their generalist character and may be due to the high rate of environment degradation revealed by the regional landscape's mosaic aspect.

Applying the Jackknife Method to Significance Tests of Diagnostic Diversity

The sample theory of normal diversity indices is complex. Distributionfree methods, such as the jackknife method, can easily be used to determine confidence intervals and testing diversity. Jackknife estimates and their variances for a number of different diversity indices are described in this paper. A simple numerical example is given for demonstrating this method. Discrimination based on confidence intervals is also discussed. It is assumed that there is a special correlation between the sensitivity parameter m and the relative width of confidence intervals in the Hurlbert index family. It is shown that the usual estimation of the Hurlbert index coincides with the relating jackknife estimate. For demonstration, diagnoses registered in a set of death certificates are used. There is a considerable diversity in diagnoses among different diagnostic groups: the diversity is largest in autopsy reports, whereas it is non-significant in GP’s reports and in reports of physicians authorized to issue death certificates. Knowing that autopsy reports tend to be fairly accurate, our research findings seem to confirm the hypothesis that there is a correlation between reliability and diversity of diagnoses.