Tent Template Functions Vectorization of Persistent Homology — step_vpd_tent_template

The function step_vpd_tent_template_functions() creates a specification of a recipe step that will convert a list-column of 3-column matrices of persistence data to a list-column of 1-row matrices of vectorizations.

Usage

step_vpd_tent_template_functions(
  recipe,
  ...,
  role = "predictor",
  trained = FALSE,
  hom_degree = 0L,
  tent_size = NULL,
  num_bins = 10L,
  tent_shift = NULL,
  columns = NULL,
  keep_original_cols = TRUE,
  skip = FALSE,
  id = rand_id("vpd_tent_template_functions")
)

Arguments

recipe: A recipe object. The step will be added to the sequence of operations for this recipe.
...: One or more selector functions to choose variables for this step. See selections() for more details.
role: For model terms created by this step, what analysis role should they be assigned? By default, the new columns created by this step from the original variables will be used as predictors in a model.
trained: A logical to indicate if the quantities for preprocessing have been estimated.
hom_degree: The homological degree of the features to be transformed.
tent_size: The length of the increment used to discretize tent template functions.
num_bins: The number of bins along each axis in the discretization grid.
tent_shift: The vertical shift applied to the discretization grid.
columns: A character string of the selected variable names. This field is a placeholder and will be populated once prep() is used.
keep_original_cols: A logical to keep the original variables in the output. Defaults to FALSE.
skip: A logical. Should the step be skipped when the recipe is baked by bake()? While all operations are baked when prep() is run, some operations may not be able to be conducted on new data (e.g. processing the outcome variable(s)). Care should be taken when using skip = TRUE as it may affect the computations for subsequent operations.
id: A character string that is unique to this step to identify it.

Value

An updated version of recipe with the new step added to the sequence of any existing operations.

Details

Persistent homology is usually encoded as birth–death pairs (barcodes or diagrams), but the space of persistence data sets does not satisfy convenient statistical properties. Such applications as hypothesis testing and machine learning benefit from transformations of persistence data, often to Hilbert spaces (vector spaces with inner products and induced metrics).

Engine

The tent template functions vectorization deploys TDAvec::computeTemplateFunction(). See there for definitions and references.

Tuning Parameters

This step has 4 tuning parameters:

hom_degree: Homological degree (type: integer, default: 0L)
tent_size: Discretization grid increment (type: double, default: NULL)
num_bins: Discretization grid bins (type: integer, default: 10L)
tent_shift: Discretization grid shift (type: double, default: NULL)

Examples

library(recipes)

# inspect vectorized features
volc_dat <- data.frame(image = I(list(volcano / 10)))
recipe(~ image, data = volc_dat) %>% 
  step_pd_raster(image, method = "link_join") %>% 
  step_vpd_tent_template_functions(image, hom_degree = 1) %>% 
  print() -> volc_rec
#> 
#> ── Recipe ──────────────────────────────────────────────────────────────────────
#> 
#> ── Inputs 
#> Number of variables by role
#> predictor: 1
#> 
#> ── Operations 
#> • persistent features from a cubical filtration of: image
#> • tent template functions of: image
print(volc_rec)
#> 
#> ── Recipe ──────────────────────────────────────────────────────────────────────
#> 
#> ── Inputs 
#> Number of variables by role
#> predictor: 1
#> 
#> ── Operations 
#> • persistent features from a cubical filtration of: image
#> • tent template functions of: image
volc_rec %>% 
  prep(training = volc_dat) %>% 
  bake(new_data = volc_dat)
#> # A tibble: 1 × 111
#>   image  image_tf_1 image_tf_2 image_tf_3 image_tf_4 image_tf_5 image_tf_6
#>   <list>      <dbl>      <dbl>      <dbl>      <dbl>      <dbl>      <dbl>
#> 1 <PHom>          0          0          0          0          0          0
#> # ℹ 104 more variables: image_tf_7 <dbl>, image_tf_8 <dbl>, image_tf_9 <dbl>,
#> #   image_tf_10 <dbl>, image_tf_11 <dbl>, image_tf_12 <dbl>, image_tf_13 <dbl>,
#> #   image_tf_14 <dbl>, image_tf_15 <dbl>, image_tf_16 <dbl>, image_tf_17 <dbl>,
#> #   image_tf_18 <dbl>, image_tf_19 <dbl>, image_tf_20 <dbl>, image_tf_21 <dbl>,
#> #   image_tf_22 <dbl>, image_tf_23 <dbl>, image_tf_24 <dbl>, image_tf_25 <dbl>,
#> #   image_tf_26 <dbl>, image_tf_27 <dbl>, image_tf_28 <dbl>, image_tf_29 <dbl>,
#> #   image_tf_30 <dbl>, image_tf_31 <dbl>, image_tf_32 <dbl>, …

# dimension-reduce using vectorized features
data(permeability_qsar, package = "modeldata")
permeability_qsar %>% 
  transform(perm_cut = cut(permeability, breaks = seq(0, 60, 10))) %>% 
  subset(select = -permeability) %>% 
  tidyr::nest(chem_fp = -perm_cut) %>% 
  print() -> perm_dat
#> # A tibble: 6 × 2
#>   perm_cut chem_fp               
#>   <fct>    <list>                
#> 1 (10,20]  <tibble [20 × 1,107]> 
#> 2 (0,10]   <tibble [110 × 1,107]>
#> 3 (20,30]  <tibble [7 × 1,107]>  
#> 4 (30,40]  <tibble [8 × 1,107]>  
#> 5 (40,50]  <tibble [16 × 1,107]> 
#> 6 (50,60]  <tibble [4 × 1,107]>  
recipe(perm_cut ~ chem_fp, data = perm_dat) %>% 
  step_pd_point_cloud(chem_fp, max_hom_degree = 2) %>% 
  step_vpd_tent_template_functions(chem_fp, hom_degree = 1) %>% 
  step_pca(starts_with("chem_fp_"), num_comp = 2) %>%
  print() -> perm_rec
#> 
#> ── Recipe ──────────────────────────────────────────────────────────────────────
#> 
#> ── Inputs 
#> Number of variables by role
#> outcome:   1
#> predictor: 1
#> 
#> ── Operations 
#> • persistent features from a Rips filtration of: chem_fp
#> • tent template functions of: chem_fp
#> • PCA extraction with: starts_with("chem_fp_")
perm_est <- prep(perm_rec, training = perm_dat)
#> Warning: no non-missing arguments to min; returning Inf
#> Warning: no non-missing arguments to max; returning -Inf
#> Warning: no non-missing arguments to min; returning Inf
#> Warning: no non-missing arguments to max; returning -Inf
#> Warning: no non-missing arguments to min; returning Inf
#> Warning: no non-missing arguments to max; returning -Inf
#> Warning: no non-missing arguments to min; returning Inf
#> Warning: no non-missing arguments to max; returning -Inf
#> Warning: no non-missing arguments to min; returning Inf
#> Warning: no non-missing arguments to max; returning -Inf
#> Warning: no non-missing arguments to min; returning Inf
#> Warning: no non-missing arguments to max; returning -Inf
perm_res <- bake(perm_est, new_data = perm_dat)
# inspect results
tidy(perm_rec)
#> # A tibble: 3 × 6
#>   number operation type                        trained skip  id                 
#>    <int> <chr>     <chr>                       <lgl>   <lgl> <chr>              
#> 1      1 step      pd_point_cloud              FALSE   FALSE pd_point_cloud_pjw…
#> 2      2 step      vpd_tent_template_functions FALSE   FALSE vpd_tent_template_…
#> 3      3 step      pca                         FALSE   FALSE pca_WcEgZ          
tidy(perm_rec, number = 2)
#> # A tibble: 1 × 3
#>   terms   value id                               
#>   <chr>   <dbl> <chr>                            
#> 1 chem_fp    NA vpd_tent_template_functions_WjKWW
tidy(perm_est, number = 2)
#> # A tibble: 1 × 3
#>   terms   value id                               
#>   <chr>   <dbl> <chr>                            
#> 1 chem_fp    NA vpd_tent_template_functions_WjKWW
# visualize results
with(perm_res, {
  plot(PC1, PC2, type = "n", asp = 1)
  text(PC1, PC2, labels = perm_cut)
})