constructive_aggregation-class

Constructive aggregation via out-performance contiguity
This method denotes the idea of rearranging a portfolio of
models (base_ensemble) into different overlapping
subsets. These subsets are aggregated (aggregate_subsets)
into combined opinions, forming new models. These models
are combined into a final decision through aggregate_hypos.

internal

A framework for dynamically combining forecasting models for time series forecasting predictive tasks. It leverages machine learning models from other packages to automatically combine expert advice using metalearning and other state-of-the-art forecasting combination approaches. The predictive methods receive a data matrix as input, representing an embedded time series, and return a predictive ensemble model. The ensemble use generic functions 'predict()' and 'forecast()' to forecast future values of the time series. Moreover, an ensemble can be updated using methods, such as 'update_weights()' or 'update_base_models()'. A complete description of the methods can be found in: Cerqueira, V., Torgo, L., Pinto, F., and Soares, C. "Arbitrated Ensemble for Time Series Forecasting." to appear at: Joint European Conference on Machine Learning and Knowledge Discovery in Databases. Springer International Publishing, 2017; and Cerqueira, V., Torgo, L., and Soares, C.: "Arbitrated Ensemble for Solar Radiation Forecasting." International Work-Conference on Artificial Neural Networks. Springer, 2017 <doi:10.1007/978-3-319-59153-7_62>.

Vitor Cerqueira

tsensembler

Dynamic Ensembles for Time Series Forecasting

Luis Torgo

Carlos Soares

constructive_aggregation-class function

<dl class="dl-horizontal">
<dt><code>base_ensemble</code></dt><dd>object of class <code><a rd-options="" href="/link/base_ensemble-class?package=tsensembler&version=0.0.5" data-mini-rdoc="tsensembler::base_ensemble-class">base_ensemble-class</a></code>.
It contains the base models used that can be used for predicting
new data or forecasting future values;</dd></dl><dt><code>committee_set</code></dt><dd>a list of ids -- the individual models
of each subset;</dd><dt><code>form</code></dt><dd>formula;</dd><dt><code>specs</code></dt><dd>object of class <code><a rd-options="" href="/link/model_specs-class?package=tsensembler&version=0.0.5" data-mini-rdoc="tsensembler::model_specs-class">model_specs-class</a></code>. Contains
the parameter setting information for training the
base models;</dd><dt><code>lambda</code></dt><dd>window size used to average loss. How far to
to back in time.</dd><dt><code>alpha</code></dt><dd>contiguity size. How long should a subset
outperform other for it to be considered in the committee</dd><dt><code>recent_series</code></dt><dd>the most recent <code>lambda</code> observations.</dd><dt><code>out_of_bag</code></dt><dd>Out of bag observations used to compute
the subsets</dd><dt><code>aggregate_subsets</code></dt><dd>aggregation approach for the set of subsets.</dd><dt><code>aggregate_hypos</code></dt><dd>final aggregation approach. How should the
combined opinions be aggregated.</dd>

constructive_aggregation-class: constructive_aggregation-class

Description

Arguments

Slots