Machine learning has recently entered the mortality literature in order to improve the forecasts of stochastic mortality models. This paper proposes to use two pure, tree-based machine learning models: random forests and gradient boosting, based on the differenced log-mortality rates to produce more accurate mortality forecasts. These forecasts are compared with forecasts from traditional, stochastic mortality models and with forecasts from random forests and gradient boosting variants of the stochastic models. The comparisons are based on the Model Confidence Set procedure. The results show that the pure, tree-based models significantly outperform all other models in the majority of cases considered. To address the lack of interpretability issue associated with machine learning models, we demonstrate how to extract information about the relationships uncovered by the tree-based models. For this purpose, we consider variable importance, partial dependence plots, and variable split conditions. Results from the in-sample fit suggest that tree-based models can be very useful tools for detecting patterns within and between variables that are not commonly identifiable with traditional methods.

In this study, prediction of aircraft Estimated Time of Arrival (ETA) is proposed using machine learning algorithms. Accurate prediction of ETA is important for management of delay and air traffic flow, runway assignment, gate assignment, collaborative decision making (CDM), coordination of ground personnel and equipment, and optimisation of arrival sequence etc. Machine learning is able to learn from experience and make predictions with weak assumptions or no assumptions at all. In the proposed approach, general flight information, trajectory data and weather data were obtained from different sources in various formats. Raw data were converted to tidy data and inserted into a relational database. To obtain the features for training the machine learning models, the data were explored, cleaned and transformed into convenient features. New features were also derived from the available data. Random forests and deep neural networks were used to train the machine learning models. Both models can predict the ETA with a mean absolute error (MAE) less than 6min after departure, and less than 3min after terminal manoeuvring area (TMA) entrance. Additionally, a web application was developed to dynamically predict the ETA using proposed models.

We study the scaling limit of a random forest with prescribed degree sequence in the regime that the largest tree consists of all but a vanishing fraction of nodes. We give a description of the limit of the forest consisting of the small trees, by relating a plane forest to a marked cyclic forest and its corresponding skip-free walk.

The panpipe is a musical instrument composed of end-blown tubes of different lengths tied together. They can be traced back to the Neolithic, and they have been found at prehistoric sites in China, Europe and South America. Panpipes display substantial variation in space and time across functional and aesthetic dimensions. Finding similarities in panpipes that belong to distant human groups poses a challenge to cultural evolution: while some have claimed that their relative simplicity speaks for independent inventions, others argue that strong similarities of specific features in panpipes from Asia, Oceania and South America suggest long-distance diffusion events. We examined 20 features of a worldwide sample of 401 panpipes and analysed statistically whether instrument features can successfully be used to determine provenance. The model predictions suggest that panpipes are reliable provenance markers, but we found an unusual classification error in which Melanesian panpipes are predicted as originating in South America. Although this pattern may be signalling a diffusion event, other factors such as convergence and preservation biases may play a role. Our analyses show the potential of cultural evolution research on music that incorporates material evidence, which in this study includes both archaeological and ethnographic samples preserved in museum collections.

Statistical methods are usually applied in examining diet–disease associations, whereas factor analysis is commonly used for dietary pattern recognition. Recently, machine learning (ML) has been also proposed as an alternative technique in health classification. In this work, the predictive accuracy of statistical v. ML methodologies as regards the association of dietary patterns on CVD risk was tested. During 2001–2002, 3042 men and women (45 (sd 14) years) were enrolled in the ATTICA study. In 2011–2012, the 10-year CVD follow-up was performed among 2020 participants. Item Response Theory was applied to create a metric of combined 10-year cardiometabolic risk, the ‘Cardiometabolic Health Score’, that incorporated incidence of CVD, diabetes, hypertension and hypercholesterolaemia. Factor analysis was performed to extract dietary patterns, on the basis of either foods or nutrients consumed; linear regression analysis was used to assess their association with the cardiometabolic score. Two ML techniques (k-nearest-neighbor’s algorithm and random-forests decision tree) were applied to evaluate participants’ health based on dietary information. Factor analysis revealed five and three factors from foods and nutrients, respectively, explaining 54 and 65 % of the total variation in intake. Nutrient and food pattern regression models showed similar accuracy in correctly classifying an individual according to the cardiometabolic risk (R2=9·6 % and R2=8·3 %, respectively). ML techniques were superior compared with linear regression in correct classification of the individuals according to the Health Score (accuracy approximately 38 v. 6 %, respectively), whereas the two ML methods showed equal classification ability. Conclusively, ML methods could be a valuable tool in the field of nutritional epidemiology, leading to more accurate disease-risk evaluation.

Objective. We used data on fleas and gamasid mites parasitic on 8 species of small mammals to test whether (a) species composition of ectoparasite infracommunities may be used to predict host habitat occupancy and (b) the accuracy of this prediction differs between ectoparasite taxa, host genders and seasons. Methods. We used a Random Forests algorithm that is based on the methodology of classification trees. Results. The accuracy of prediction of habitat occupancy was relatively low and varied substantially among host species. The combined rate of the correct prediction of host habitat occupancy from data on ectoparasites was significantly higher than 50%, albeit being relatively low. The accuracy of prediction (a) did not differ between male and female hosts when it was based on species composition of fleas in summer or of mites in summer and winter, (b) was significantly higher in male hosts than in female hosts when the winter data on fleas were used and (c) was significantly higher for flea than mite assemblages. The effect of season was found in mites but not in fleas with the accuracy of prediction being significantly higher in summer than in winter assemblages. Conclusions. Ectoparasites appeared to be not especially useful as biological markers for distinguishing host populations in different habitats in temperate zones.

We address the effect of spatial scale and temporal variation on model generality when forming predictive models for fish assignment using a new data mining approach, Random Forests (RF), to variable biological markers (parasite community data). Models were implemented for a fish host-parasite system sampled along the Mediterranean and Atlantic coasts of Spain and were validated using independent datasets. We considered 2 basic classification problems in evaluating the importance of variations in parasite infracommunities for assignment of individual fish to their populations of origin: multiclass (2–5 population models, using 2 seasonal replicates from each of the populations) and 2-class task (using 4 seasonal replicates from 1 Atlantic and 1 Mediterranean population each). The main results are that (i) RF are well suited for multiclass population assignment using parasite communities in non-migratory fish; (ii) RF provide an efficient means for model cross-validation on the baseline data and this allows sample size limitations in parasite tag studies to be tackled effectively; (iii) the performance of RF is dependent on the complexity and spatial extent/configuration of the problem; and (iv) the development of predictive models is strongly influenced by seasonal change and this stresses the importance of both temporal replication and model validation in parasite tagging studies.