None
ts_random_walk_plot() - Creates a side-by-side faceted plot
for random walk simulations showing both the random variable and the random walk
path (cumulative product).ts_ma_plot() to use ggplot2::facet_wrap() instead of
cowplot::plot_grid() for stacking plots. The function now returns only ggplot2
output and no longer includes xts plotting functionality. The return value has been
simplified from 6 items to 2 items: pgrid (ggplot2 facet_wrap plot object) and
data_summary_tbl (summary data table). Removed return values: data_trans_xts,
data_diff_xts_a, data_diff_xts_b, and xts_plt.None
None
None
ts_random_walk() that would generate 3 runs when run was
set to less than 2.timetk::step_holiday_signature() as it is throwing recycle erros in
boilerplate functions and I currently cannot figure out why.None
util_log_ts()util_singlediff_ts()util_doublediff_ts()util_difflog_ts()util_doubledifflog_ts()ts_growth_rate_vec()auto_stationarize()ts_auto_arima() to utilize the parsnip engine of auto_arima
if .tune is set to FALSENone
ts_growth_rate_vec()ts_adf_test()auto_stationarize()ts_growth_rate_augment().true param to FALSENone
None
None
None
tune::show_best(n = 1)
instead of Inf and using dplyr::slice(1)None
ts_ma_plt() errors stemming from deprecations. Also fixed
examples of all boilerplate functions.None
ts_geometric_brownian_motion()ts_brownian_motion_augment()ts_geometric_brownian_motion_augment()ts_brownian_motion_plot()ts_brownian_motion() 49x speedup by way of
vectorization..motion_typets_vva_plot()None
ts_brownian_motion()ts_scedacity_scatter_plot()None
None
ts_lag_correlation() to fix a bug in the correlation matrix calculation where columns may come through that are not numeric and are not part of the original value and it's lags.None
None
recipes::check_type()
on recipe functions.ts_model_spec_tune_template() to set regression as the
argument to parsnip::set_mode() which fires a failure in the ts_model_auto_tune()
not running on newer versions of parsnipNone
ts_wfs_xgboost()ts_calendar_heatmap_plot() Change weekdays and Monthls
to abbreviated labels.ts_sma_plot() There is a change in the API of this function.
It now requires a data.frame/tibble to be passed to the .data parameter, and
it also now requires the input of a date column and value column. This also now
no longer returns invisible. There was also a fix in the sliding calculation
to appropriately use the given value column.ts_extract_auto_fitted_workflow() Which will pull
out the fitted workflow from any of the Boilerplate functions.ts_auto_lm() by dropping step_rm() and step_corr() which
would prevent calibrate_and_plot() from working due to modeltime_calibration()
failing. Also dropped unused parameters from function and documentation.ts_lag_correlation() select statement.None
ts_time_event_analysis_tbl()ts_lag_correlation()ci_hi() and ci_lo()ts_event_analysis_plot()ts_model_auto_tune() and ts_model_spec_tune_template() to
accept svm_poly and svm_rbf. This helps in allowing users to auto tune models
that are create by ts_wfs_svm_poly() and ts_wfs_svm_rbf() functions respectively.
Also added "model_spec_class" to the output of the ts_model_auto_tune() function.None
ts_auto_arima() boiler plate function.color_blind() ts_scale_fill_colorblind() and
ts_scale_color_colorblind()ts_auto_smooth_es()ts_auto_theta()ts_auto_lm()ts_auto_svm_poly()ts_auto_svm_rbf()ts_auto_arima_xgboots() when .tune is FALSE.tune::tune()healthyR to healthyR.ai in
anticipation of dropping kmeans functionality from healthyRNone
ts_arima_simulator()ts_feature_cluster()ts_feature_cluster_plot()ts_auto_glmnet()ts_auto_xgboost()ts_auto_arima_xgboost()ts_auto_mars()ts_auto_exp_smoothing()ts_auto_croston()ts_auto_nnetar()ts_auto_prophet_reg()ts_auto_prophet_boost()[recipes::print_step()] method.ggplot2::theme_minimal()hardhat to DESCRIPTION since functionality like extracting
dials parameters was taken out of dials and moved to hardhat.None
ts and Fitted tibble data to output.ts and Residuals tibble data to output.Arima() models with xreg to ts_forecast_simulator()model_extraction_helper() to utilize forecast:::arima.string()
under the hood for Arima arima and auto.arima models produced by the forecast
package.crayon, cli, and rstudioapi since all it did was
make a welcome message that can be done with regular print() method.None
ts_qq_plot()ts_scedacity_scatter_plot()ts_model_rank_tbl()model_extraction_helper() to grab workflow model_spec
and model_fit objects.None
ts_vva_plot()ts_model_compare()ts_acceleration_vec()ts_acceleration_augment()step_ts_acceleration()ts_velocity_vec()ts_velocity_augment()step_ts_velocity().date_col to ts_sma_plot() so that if a tibble is passed
the appropriate column is passed to the ggplot object.model_extraction_helper() function to extract workflow fit
models.None
None
Fix #143 - Drop mtry = tune::tune() from ts_model_spec_tune_template as it
causes issues downstream.
None
tidy_fft() functionts_info_tbl() functionts_sma_plot() functionts_to_tbl() functionts_model_auto_tune() functionts_model_spec_tune_template() functionts_wfs_auto_arima() functionts_wfs_arima_boost() functionts_wfs_ets_reg() functionts_wfs_nnetar_reg() functionts_wfs_prophet_reg() functionts_auto_recipe() bug that forced the change of column names
in the output. This has been fixed and the column names supplied will now be in
the recipe terms.None
ts_forecast_simulator() functioncalibrate_and_plot() helper functionts_wfs_lin_reg(), ts_wfs_mars(), ts_wfs_svm_poly(), ts_wfs_svm_rbf()model_extraction_helper() helper functionts_ma_plot() plotting functionts_calendar_heatmap_plot() plotting functionts_splits_plot() plotting functionNone
ts_auto_recipe() functionNone
NEWS.md file to track changes to the package.