Decision trees are surprisingly adaptive in three important respects: They automatically (1) adapt to favorable conditions near the Bayes decision boundary; (2) focus on data distributed on lower dimensional manifolds; (3) reject irrelevant features. In this paper we examine a decision tree based on dyadic splits that adapts to each of these conditions to achieve minimax optimal rates of convergence. The proposed classifier is the first known to achieve these optimal rates while being practical and im- plementable.

This paper proposes a novel boosting algorithm called VadaBoost which is motivated by recent empirical Bernstein bounds. VadaBoost iteratively minimizes a cost function that balances the sample mean and the sample variance of the exponential loss. Each step of the proposed algorithm minimizes the cost efficiently by providing weighted data to a weak learner rather than requiring a brute force evaluation of all possible weak learners. Thus, the proposed algorithm solves a key limitation of previous empirical Bernstein boosting methods which required brute force enumeration of all possible weak learners. Experimental results confirm that the new algorithm achieves the performance improvements of EBBoost yet goes beyond decision stumps to handle any weak learner.

This paper introduces timeline trees, which are partial models of partially observable environments. Timeline trees are given some specific predictions to make and learn a decision tree over history. The main idea of timeline trees is to use temporally abstract features to identify and split on features of key events, spread arbitrarily far apart in the past (whereas previous decision-tree-based methods have been limited to a finite suffix of history). Experiments demonstrate that timeline trees can learn to make high quality predictions in complex, partially observable environments with high-dimensional observations (e.g. an arcade game).

This study is devoted to two of the oldest known manuscripts in which the oeuvre of the medieval mystical author Hadewijch has been preserved: Brussels, KBR, 2879-2880 (ms. On the basis of codicological and contextual arguments, it is assumed that the scribe who produced B used A as an exemplar. While the similarities in both layout and content between the two manuscripts are striking, the present article seeks to identify the differences. After all, regardless of the intention to produce a copy that closely follows the exemplar, subtle linguistic variation is apparent. Divergences relate to spelling conventions, but also to the way in which words are abbreviated (and the extent to which abbreviations occur). The present study investigates the spelling profiles of the scribes who produced mss. In the first part of this study, we will present both manuscripts in more detail, after which we will consider prior research carried out on scribal profiling. The current study both builds and expands on Kestemont (2015). Next, we outline the methodology used to analyse and measure the degree of scribal appropriation that took place when ms. B was copied off the exemplar ms. A. After this, we will discuss the results obtained, focusing on the scribal variation that can be found both at the level of individual words and n-grams. To this end, we use machine learning to identify the most distinctive features that separate manuscript A from B. Finally, we look at possible diachronic trends in the appropriation by B's scribe of his exemplar. We argue that scribal takeovers in the exemplar impacts the practice of the copying scribe, while transitions to a different content matter cause little to no effect. INTRODUCTION Among the Royal Library of Belgium's (KBR) extraordinarily rich collection are two fourteenth-century manuscripts that are of great importance to the field of medieval Dutch literature in general, and that of mysticism in the Low Countries in particular: KBR 2879-80 and KBR 2877-78. Both manuscripts contain the complete oeuvre - consisting of letters, visions, songs, and poems - of the mystical writer Hadewijch. Since unambiguous biographical data are lacking, the historical figure of Hadewijch is largely shrouded in mystery. Through her work, however, one can get a modest glimpse of who she was and when she lived. Researchers who undertook this quest situate Hadewijch in the religious women's movement (mulieres religiosae) of the thirteenth century [Mommaers, 2003; Fraeters & Willaert, 2009, p. 13-19; Fraeters, 2013; Willaert, 2013].

Random forest is a popular machine learning approach for the analysis of high-dimensional data because it is flexible and provides variable importance measures for the selection of relevant features. However, the complex relationships between the features are usually not considered for the selection and thus also neglected for the characterization of the analysed samples. Here we propose two novel approaches that focus on the mutual impact of features in random forests. Mutual forest impact (MFI) is a relation parameter that evaluates the mutual association of the featurs to the outcome and, hence, goes beyond the analysis of correlation coefficients. Mutual impurity reduction (MIR) is an importance measure that combines this relation parameter with the importance of the individual features. MIR and MFI are implemented together with testing procedures that generate p-values for the selection of related and important features. Applications to various simulated data sets and the comparison to other methods for feature selection and relation analysis show that MFI and MIR are very promising to shed light on the complex relationships between features and outcome. In addition, they are not affected by common biases, e.g. that features with many possible splits or high minor allele frequencies are prefered.

In this note, we establish the hardness of approximation of the problem of computing the minimal size of a $\delta$-sufficient reason for decision trees.

Due to rapidly rising healthcare costs worldwide, there is significant interest in controlling them. An important aspect concerns price transparency, as preliminary efforts have demonstrated that patients will shop for lower costs, driving efficiency. This requires the data to be made available, and models that can predict healthcare costs for a wide range of patient demographics and conditions. We present an approach to this problem by developing a predictive model using machine-learning techniques. We analyzed de-identified patient data from New York State SPARCS (statewide planning and research cooperative system), consisting of 2.3 million records in 2016. We built models to predict costs from patient diagnoses and demographics. We investigated two model classes consisting of sparse regression and decision trees. We obtained the best performance by using a decision tree with depth 10. We obtained an R-square value of 0.76 which is better than the values reported in the literature for similar problems.

The recent boom in AI has clearly shown the power of deep neural networks in various tasks, especially in the field of classification problems where the data is high-dimensional and has complex, non-linear relationships with the target variables. However, explaining the decisions of any neural classifier is an incredibly hard problem. While many post-hoc methods such as DeepLift [2] and Layer-Wise Relevance Propagation [3] can help with explaining individual decisions, explaining the global decision mechanisms (or what the model generally looks for) is much more difficult. Because of this, many practitioners in high-stakes fields instead opt for more interpretable models like basic Decision Trees since the decision hierarchy can be clearly visualized and understood by stakeholders. However, basic trees by themselves often do not provide enough accuracy for the task at hand and often ensemble methods like Bagging or Boosting are used to improve the model's performance.

Stroke is the second leading cause of death worldwide. Machine learning classification algorithms have been widely adopted for stroke prediction. However, these algorithms were evaluated using different datasets and evaluation metrics. Moreover, there is no comprehensive framework for stroke data analytics. This paper proposes an intelligent stroke prediction framework based on a critical examination of machine learning prediction algorithms in the literature. The five most used machine learning algorithms for stroke prediction are evaluated using a unified setup for objective comparison. Comparative analysis and numerical results reveal that the Random Forest algorithm is best suited for stroke prediction.

The decision to choose between Random Forest and Decision Tree models depends on the complexity of the problem, the size of the dataset, the interpretability of the model, and the trade-off between accuracy and computational efficiency. Complexity of the problem: Decision trees are simpler and easier to interpret, making them a good choice for smaller and less complex problems. However, for larger and more complex problems, Random Forest models can provide better accuracy due to their ability to combine multiple decision trees. Size of the dataset: Decision trees can be sensitive to noise and outliers, and may overfit the data if the dataset is too small. Random Forest models can be more robust to noise and overfitting, making them a good choice for smaller datasets.