In this work, we introduce the Personalized Online Super Learner (POSL) -- an online ensembling algorithm for streaming data whose optimization procedure accommodates varying degrees of personalization. Namely, POSL optimizes predictions with respect to baseline covariates, so personalization can vary from completely individualized (i.e., optimization with respect to baseline covariate subject ID) to many individuals (i.e., optimization with respect to common baseline covariates). As an online algorithm, POSL learns in real-time. POSL can leverage a diversity of candidate algorithms, including online algorithms with different training and update times, fixed algorithms that are never updated during the procedure, pooled algorithms that learn from many individuals' time-series, and individualized algorithms that learn from within a single time-series. POSL's ensembling of this hybrid of base learning strategies depends on the amount of data collected, the stationarity of the time-series, and the mutual characteristics of a group of time-series. In essence, POSL decides whether to learn across samples, through time, or both, based on the underlying (unknown) structure in the data. For a wide range of simulations that reflect realistic forecasting scenarios, and in a medical data application, we examine the performance of POSL relative to other current ensembling and online learning methods. We show that POSL is able to provide reliable predictions for time-series data and adjust to changing data-generating environments. We further cultivate POSL's practicality by extending it to settings where time-series enter/exit dynamically over chronological time.

We investigate the fairness concerns of training a machine learning model using data with missing values. Even though there are a number of fairness intervention methods in the literature, most of them require a complete training set as input. In practice, data can have missing values, and data missing patterns can depend on group attributes (e.g. gender or race). Simply applying off-the-shelf fair learning algorithms to an imputed dataset may lead to an unfair model. In this paper, we first theoretically analyze different sources of discrimination risks when training with an imputed dataset. Then, we propose an integrated approach based on decision trees that does not require a separate process of imputation and learning. Instead, we train a tree with missing incorporated as attribute (MIA), which does not require explicit imputation, and we optimize a fairness-regularized objective function. We demonstrate that our approach outperforms existing fairness intervention methods applied to an imputed dataset, through several experiments on real-world datasets.

Classification of time series signals has become an important construct and has many practical applications. With existing classifiers we may be able to accurately classify signals, however that accuracy may decline if using a reduced number of attributes. Transforming the data then undertaking reduction in dimensionality may improve the quality of the data analysis, decrease time required for classification and simplify models. We propose an approach, which chooses suitable wavelets to transform the data, then combines the output from these transforms to construct a dataset to then apply ensemble classifiers to. We demonstrate this on different data sets, across different classifiers and use differing evaluation methods. Our experimental results demonstrate the effectiveness of the proposed technique, compared to the approaches that use either raw signal data or a single wavelet transform.

Prompt severity assessment model of confirmed patients who were infected with infectious diseases could enable efficient diagnosis and alleviate the burden on the medical system. This paper provides the development processes of the severity assessment model using machine learning techniques and its application on SARS-CoV-2 patients. Here, we highlight that our model only requires basic patients' basic personal data, allowing for them to judge their own severity. We selected the boosting-based decision tree model as a classifier and interpreted mortality as a probability score after modeling. Specifically, hyperparameters that determine the structure of the tree model were tuned using the Bayesian optimization technique without any knowledge of medical information. As a result, we measured model performance and identified the variables affecting the severity through the model. Finally, we aim to establish a medical system that allows patients to check their own severity and informs them to visit the appropriate clinic center based on the past treatment details of other patients with similar severity.

Symbolic learning represents the most straightforward approach to interpretable modeling, but its applications have been hampered by a single structural design choice: the adoption of propositional logic as the underlying language. Recently, more-than-propositional symbolic learning methods have started to appear, in particular for time-dependent data. These methods exploit the expressive power of modal temporal logics in powerful learning algorithms, such as temporal decision trees, whose classification capabilities are comparable with the best non-symbolic ones, while producing models with explicit knowledge representation. With the intent of following the same approach in the case of spatial data, in this paper we: i) present a theory of spatial decision tree learning; ii) describe a prototypical implementation of a spatial decision tree learning algorithm based, and strictly extending, the classical C4.5 algorithm; and iii) perform a series of experiments in which we compare the predicting power of spatial decision trees with that of classical propositional decision trees in several versions, for a multi-class image classification problem, on publicly available datasets. Our results are encouraging, showing clear improvements in the performances from the propositional to the spatial models, which in turn show higher levels of interpretability.

We introduce WildWood (WW), a new ensemble algorithm for supervised learning of Random Forest (RF) type. While standard RF algorithms use bootstrap out-of-bag samples to compute out-of-bag scores, WW uses these samples to produce improved predictions given by an aggregation of the predictions of all possible subtrees of each fully grown tree in the forest. This is achieved by aggregation with exponential weights computed over out-of-bag samples, that are computed exactly and very efficiently thanks to an algorithm called context tree weighting. This improvement, combined with a histogram strategy to accelerate split finding, makes WW fast and competitive compared with other well-established ensemble methods, such as standard RF and extreme gradient boosting algorithms.

Decisions and the underlying rules are indispensable for driving process execution during runtime, i.e., for routing process instances at alternative branches based on the values of process data. Decision rules can comprise unary data conditions, e.g., age > 40, binary data conditions where the relation between two or more variables is relevant, e.g. temperature1 < temperature2, and more complex conditions that refer to, for example, parts of a medical image. Decision discovery aims at automatically deriving decision rules from process event logs. Existing approaches focus on the discovery of unary, or in some instances binary data conditions. The discovered decision rules are usually evaluated using accuracy, but not with regards to their semantics and meaningfulness, although this is crucial for validation and the subsequent implementation/adaptation of the decision rules. Hence, this paper compares three decision mining approaches, i.e., two existing ones and one newly described approach, with respect to the meaningfulness of their results. For comparison, we use one synthetic data set for a realistic manufacturing case and the two real-world BPIC 2017/2020 logs. The discovered rules are discussed with regards to their semantics and meaningfulness.

Knowledge transfer from a complex high performing model to a simpler and potentially low performing one in order to enhance its performance has been of great interest over the last few years as it finds applications in important problems such as explainable artificial intelligence, model compression, robust model building and learning from small data. Known approaches to this problem (viz. Knowledge Distillation, Model compression, ProfWeight, etc.) typically transfer information directly (i.e. in a single/one hop) from the complex model to the chosen simple model through schemes that modify the target or reweight training examples on which the simple model is trained. In this paper, we propose a meta-approach where we transfer information from the complex model to the simple model by dynamically selecting and/or constructing a sequence of intermediate models of decreasing complexity that are less intricate than the original complex model. Our approach can transfer information between consecutive models in the sequence using any of the previously mentioned approaches as well as work in 1-hop fashion, thus generalizing these approaches. In the experiments on real data, we observe that we get consistent gains for different choices of models over 1-hop, which on average is more than 2\% and reaches up to 8\% in a particular case. We also empirically analyze conditions under which the multi-hop approach is likely to be beneficial over the traditional 1-hop approach, and report other interesting insights. To the best of our knowledge, this is the first work that proposes such a multi-hop approach to perform knowledge transfer given a single high performing complex model, making it in our opinion, an important methodological contribution.

Ensemble Learning is taking the predictions of multiple models and assume the output to be having the most votes. When you train multiple Decision Trees each on some random sampling of the dataset and for predictions you take predictions of all the trees, the output class would be the class which gets the most votes. This approach is called Random Forest. Voting classifier is when you train the data on multiple classifier such as Logistic Regression, SVM, RF and other classifiers and the majority vote is the predicted output class ie hard classifier. Voting can also be taken as soft by taking argmax of the outputs.

As a lifelong computational scientist (and now data scientist) I have always been fascinated with numbers, especially lists and tables of things ( databases!). For example, I thought early in life that I wanted to be a Math major in college and study number theory so that I could learn all of the amazing ways to do cool stuff with numbers. But I also wanted to study the wonders of the Universe as an astronomer, so I went on to get a PhD in astrophysics! That career path allowed me to study and apply all of the disciplines that I enjoy: math, physics, astronomy, computational modeling, and data science! It was numbers all the time!