tsensembler0.1.0 package

Dynamic Ensembles for Time Series Forecasting

sequential_reweighting

Sequential Re-weighting for controlling predictions' redundancy

sliding_similarity

Sliding similarity via Pearson's correlation

soft.completion

Soft Imputation

softmax

Computing the softmax

split_by

Splitting expressions by pattern

train_ade

Training procedure of for ADE

train_ade_quick

ADE training poor version Train meta-models in the training data, as o...

tsensembler

Dynamic Ensembles for Time Series Forecasting

ADE-class

Arbitrated Dynamic Ensemble

ADE

Arbitrated Dynamic Ensemble

ade_hat-class

Predictions by an ADE ensemble

ade_hat

Predictions by an ADE ensemble

ae

Computing the absolute error

base_ensemble-class

base_ensemble-class

base_ensemble

base_ensemble

base_models_loss

Computing the error of base models

best_mvr

Get best PLS/PCR model

blocked_prequential

Prequential Procedure in Blocks

bm_cubist

Fit Cubist models (M5)

bm_ffnn

Fit Feedforward Neural Networks models

bm_gaussianprocess

Fit Gaussian Process models

bm_gbm

Fit Generalized Boosted Regression models

bm_glm

Fit Generalized Linear Models

bm_mars

Fit Multivariate Adaptive Regression Splines models

bm_pls_pcr

Fit PLS/PCR regression models

bm_ppr

Fit Projection Pursuit Regression models

bm_randomforest

Fit Random Forest models

bm_svr

Fit Support Vector Regression models

bm_xgb

Base model for XGBoost

build_base_ensemble

Wrapper for creating an ensemble

build_committee

Building a committee for an ADE model

combine_predictions

Combining the predictions of several models

compute_predictions

Compute the predictions of base models

DETS-class

Dynamic Ensemble for Time Series

DETS

Dynamic Ensemble for Time Series

dets_hat-class

Predictions by an DETS ensemble

dets_hat

Predictions by an DETS ensemble

EMASE

Weighting Base Models by their Moving Average Squared Error

embed_timeseries

Embedding a Time Series

get_target

Get the target from a formula

get_top_models

Extract top learners from their weights

get_y

Get the response values from a data matrix

holdout

Holdout

intraining_estimations

Out-of-bag loss estimations

intraining_predictions

Out-of-bag predictions

l1apply

Applying lapply on the rows

learning_base_models

Training the base models of an ensemble

loss_meta_learn

Training an arbiter

meta_cubist

Training a RBR arbiter

meta_cubist_predict

Arbiter predictions via Cubist

meta_ffnn

Training a Gaussian prosadacess arbiter

meta_ffnn_predict

Arbiter predictions via linear ssmodel

meta_gp

Training a Gaussian process arbiter

meta_gp_predict

Arbiter predictions via linear model

meta_lasso

Training a LASSO arbiter

meta_lasso_predict

Arbiter predictions via linear model

meta_mars

Training a meta_mars process arbiter

meta_mars_predict

Arbiter predictions via mars model

meta_pls

Training a pls process arbiter

meta_pls_predict

Arbiter predictions via pls model

meta_ppr

Training a meta_mars process arbiter

meta_ppr_predict

Arbiter predictions via ppr model

meta_predict

Predicting loss using arbiter

meta_rf

Training a random forest arbiter

meta_rf_predict

Arbiter predictions via ranger

meta_svr

Training a Gaussian process arbiter

meta_svr_predict

Arbiter predictions via linear model

meta_xgb

Training a xgb arbiter

meta_xgb_predict

Arbiter predictions via xgb

model_recent_performance

Recent performance of models using EMASE

model_specs-class

Setup base learning models

model_specs

Setup base learning models

model_weighting

Model weighting

mse

Computing the mean squared error

normalize

Scale a numeric vector using max-min

predict-methods

Predicting new observations using an ensemble

predict_pls_pcr

predict method for pls/pcr

proportion

Computing the proportions of a numeric vector

rbind_l

rbind with do.call syntax

recent_lambda_observations

Get most recent lambda observations

rmse

Computing the root mean squared error

roll_mean_matrix

Computing the rolling mean of the columns of a matrix

se

Computing the squared error

select_best

Selecting best model according to weights

update_ade

Updating an ADE model

update_ade_meta

Updating the metalearning layer of an ADE model

update_base_models

Update the base models of an ensemble

update_weights

Updating the weights of base models

xgb_optimizer

XGB optimizer

xgb_predict

XGBoost predict function

xgb_predict_

asdasd

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>.