Modeltime workflowsets heatmap plot

generate a heatmap for each recipe and model on a object generated with the modeltime_wfs_fit() function.

modeltime_wfs_heatmap(
  .wfs_results,
  metric = "rsq",
  low_color = "#c7e9b4",
  high_color = "#253494"
)

Arguments

.wfs_results	a tibble generated with the modeltime_wfs_fit() function.
metric	a metric the metric used for the heatmap values: 'mae', 'mape','mase','smape','rmse','rsq'.
low_color	color for the worst metric (highest error or lowest rsq).
high_color	color for the better metric (lowest error or highest rsq).

Value

a ggplot heatmap.

Details

assumes that the workflows included in the 'workflow_set' object are named M_name_of_model, since the .model_id is recipe_nameMname_of_model and the 'M' is used to separate the recipe from the model name.

Examples

library(modeltime)
library(dplyr)
library(parsnip)
library(earth)

data <- sknifedatar::data_avellaneda %>% mutate(date=as.Date(date)) %>% filter(date<'2011-01-01')

recipe_date <- recipes::recipe(value ~ ., data = data) %>% 
  recipes::step_date(date, features = c('dow','doy','week','month','year')) 

mars_backward <- mars(prune_method ='backward', mode = 'regression') %>% set_engine('earth')

mars_forward <- mars(prune_method = 'forward', mode = 'regression') %>% set_engine('earth')

wfsets <- workflowsets::workflow_set(
  preproc = list(
    date = recipe_date),
  models  = list(M_mars_backward = mars_backward, 
                 M_mars_forward = mars_forward),
  cross   = TRUE)

wffits <- sknifedatar::modeltime_wfs_fit(.wfsets = wfsets, 
                                         .split_prop = 0.6, 
                                         .serie=data)
#> ── MODEL: date_M_mars_backward 
#> ✓ Training finished OK.
#> 
#> ── MODEL: date_M_mars_forward 
#> ✓ Training finished OK.
#> 
#> 
#> ── 2 models fitted ♥ ───────────────────────────────────────────────────────────
#> 
#> ── 0 models deleted x ──
#> 

sknifedatar::modeltime_wfs_heatmap(wffits, 'rsq')