Abstract:Technological advancement led to the development of tools to collect vast amounts of data usually recorded at temporal stamps or arriving over time, e.g. data from sensors. Common ways of analysing this kind of data also involve supervised classification techniques; however, despite constant improvements in the literature, learning from high-dimensional data is always a challenging task due to many issues such as, for example, dealing with the curse of dimensionality and looking for a trade-off between complexity and accuracy. Nowadays, research in functional data analysis (FDA) and statistical learning is very lively to address these drawbacks adequately. This study offers a supervised classification strategy that combines FDA and tree-based procedures. Specifically, we introduce functional classification trees, functional bagging, and functional random forest exploiting the functional principal components decomposition as a tool to extract new features and build functional classifiers. In addition, we introduce new tools to support the understanding of the classification rules, such as the functional empirical separation prototype, functional predicted separation prototype, and the leaves' functional deviance. Furthermore, we suggest some possible solutions for choosing the number of functional principal components and functional classification trees to be implemented in the supervised classification procedure. This research aims to provide an approach to improve the accuracy of the functional classifier, serve the interpretation of the functional classification rules, and overcome the classical drawbacks due to the high-dimensionality of the data. An application on a real dataset regarding daily electrical power demand shows the functioning of the supervised classification proposal. A simulation study with nine scenarios highlights the performance of this approach and compares it with other functional classification methods. The results demonstrate that this line of research is exciting and promising; indeed, in addition to the benefits of the suggested interpretative tools, we exceed the previously established accuracy records on a dataset available online.

Tree-based boosting with functional data

Boosting the Partial Least Square Algorithm for Regression Modelling

Measure inducing classification and regression trees for functional data

Boosting Functional Regression Models with FDboost

Feature Extraction in Signal Regression: A Boosting Technique for Functional Data Regression

Enriched Functional Tree-Based Classifiers: A Novel Approach Leveraging Derivatives and Geometric Features

Supervised classification of curves via a combined use of functional data analysis and tree-based methods

Robust Functional Regression with Discretely Sampled Predictors

Multivariate Boosted Trees and Applications to Forecasting and Control

BOOSTED NONPARAMETRIC HAZARDS WITH TIME-DEPENDENT COVARIATES

Unbiased recursive decision tree for supervised functional data classification with applying on electrocardiogram signals

Finding structure in data using multivariate tree boosting

Dynamic Function-on-Scalars Regression

metboost: Exploratory regression analysis with hierarchically clustered data

Tree-Structured Boosting: Connections Between Gradient Boosted Stumps and Full Decision Trees

Functional Linear Regression with Mixed Predictors

Modern Multiple Imputation with Functional Data

Partially Functional Linear Regression in High Dimensions

Robust boosting for regression problems

Functional Sufficient Dimension Reduction for Functional Data Classification

Semi-parametric forecasts of the implied volatility surface using regression trees