Package 'measure'

Description

Kuhn and Johnson (2013) used these two data sets to model the glucose yeild in large- and small-scale bioreactors:

Details

• Fifteen small-scale (5 liters) bioreactors were seeded with cells and were monitored daily for 14 days.

• Three large-scale bioreactors were also seeded with cells from the same batch and monitored daily for 14 days.

Samples were collected each day from all bioreactors and glucose was measured. The goal would be to create models on the data from the more numerous small-scale bioreactors and then evaluate if these results can accurately predict what is happening in the large-scale bioreactors.

Value

Two tibbles. For each, there are 2,651 columns whose names are numbers and these are the measured assay values (and the names are the wave numbers). The numeric column glucose has the outcome data, day is the number of days in the bioreactor, the batch_id is the reactor identifier (with "L" for large and "S" for small), and batch_sample that is the ID and the day.

Source

Kuhn and Johnson (2020), Feature Engineering and Selection, Chapman and Hall/CRC . https://bookdown.org/max/FES/ and https://github.com/topepo/FES

Examples

data(glucose_bioreactors)
dim(bioreactors_small)

Description

"These data are recorded on a Tecator Infratec Food and Feed Analyzer working in the wavelength range 850 - 1050 nm by the Near Infrared Transmission (NIT) principle. Each sample contains finely chopped pure meat with different moisture, fat and protein contents.

Details

If results from these data are used in a publication we want you to mention the instrument and company name (Tecator) in the publication. In addition, please send a preprint of your article to

Karin Thente, Tecator AB, Box 70, S-263 21 Hoganas, Sweden

The data are available in the public domain with no responsibility from the original data source. The data can be redistributed as long as this permission note is attached."

"For each meat sample the data consists of a 100 channel spectrum of absorbances and the contents of moisture (water), fat and protein. The absorbance is -log10 of the transmittance measured by the spectrometer. The three contents, measured in percent, are determined by analytic chemistry."

Included here are the meats data transformed to a long format with

modeldata::meats %>%
  rowid_to_column(var = "id") %>%
  pivot_longer(cols = starts_with("x_"),
               names_to = "channel",
               values_to = "transmittance") %>%
  mutate(channel = str_extract(channel, "[:digit:]+") %>% as.integer())

Value

Examples

data(meats_long)
str(meats_long)

Description

Set package dependencies

Usage

## S3 method for class 'step_isomap'
required_pkgs(x, ...)

Arguments

Description

Fit and subtract a baseline from a measurement signal

Usage

step_baseline(
  recipe,
  ...,
  role = NA,
  trained = FALSE,
  options = NULL,
  skip = FALSE,
  id = recipes::rand_id("measure")
)

Arguments

Description

step_measure_input_long creates a specification of a recipe step that converts measures organized in a column for the analytical results (and an option column of numeric indices) into an internal format used by the package.

Usage

step_measure_input_long(
  recipe,
  ...,
  location,
  pad = FALSE,
  role = "measure",
  trained = FALSE,
  columns = NULL,
  skip = FALSE,
  id = rand_id("measure_input_long")
)

Arguments

Details

This step is designed for data in a format where there is a column for the analytical measurement (e.g., absorption, etc.) and another with the location of the value (e.g., wave number, etc.).

step_measure_input_long() will collect those data and put them into a format used internally by this package. The data structure has a row for each independent experimental unit and a nested tibble with that sample's measure (measurement and location). It assumes that there are unique combinations of the other columns in the data that define individual patterns associated with the pattern. If this is not the case, the special values might be inappropriately restructured.

The best advice is to have a column of any type that indicates the unique sample number for each measure. For example, if there are 200 values in the measure and 7 samples, the input data (in long format) should have 1,400 rows. We advise having a column with 7 unique values indicating which of the rows correspond to each sample.

Missing Data

Currently, measure assumes that there are equal numbers of values within a sample. If there are missing values in the measurements, you'll need to pad them with missing values (as opposed to an absent row in the long format). If not, an error will occur.

Tidying

When you tidy() this step, a tibble indicating which of the original columns were used to reformat the data.

See Also

Other input/output steps: step_measure_input_wide(), step_measure_output_long(), step_measure_output_wide()

Description

step_measure_input_wide creates a specification of a recipe step that converts measures organized in multiple columns into an internal format used by the package.

Usage

step_measure_input_wide(
  recipe,
  ...,
  role = "measure",
  trained = FALSE,
  columns = NULL,
  location_values = NULL,
  skip = FALSE,
  id = rand_id("measure_input_wide")
)

Arguments

Details

This step is designed for data in a format where the analytical measurements are in separate columns.

step_measure_input_wide() will collect those data and put them into a format used internally by this package. The data structure has a row for each independent experimental unit and a nested tibble with that sample's measure (measurement and location). It assumes that there are unique combinations of the other columns in the data that define individual patterns associated with the pattern. If this is not the case, the special values might be inappropriately restructured.

The best advice is to have a column of any type that indicates the unique sample number for each measure. For example, if there are 20 rows in the input data set, the columns that are not analytically measurements show have no duplicate combinations in the 20 rows.

Tidying

When you tidy() this step, a tibble indicating which of the original columns were used to reformat the data.

See Also

Other input/output steps: step_measure_input_long(), step_measure_output_long(), step_measure_output_wide()

Examples

data(meats, package = "modeldata")

# Outcome data is to the right
names(meats) %>% tail(10)

# ------------------------------------------------------------------------------
# Ingest data without adding the location (i.e. wave number) for the spectra

rec <-
  recipe(water + fat + protein ~ ., data = meats) %>%
  step_measure_input_wide(starts_with("x_")) %>%
  prep()

summary(rec)

# ------------------------------------------------------------------------------
# Ingest data without adding the location (i.e. wave number) for the spectra

# Make up some locations for the spectra's x-axis
index <- seq(1, 2, length.out = 100)

rec <-
  recipe(water + fat + protein ~ ., data = meats) %>%
  step_measure_input_wide(starts_with("x_"), location_values = index) %>%
  prep()

summary(rec)

Description

step_measure_output_long creates a specification of a recipe step that converts measures to a format with columns for the measurement and the corresponding location (i.e., "long" format).

Usage

step_measure_output_long(
  recipe,
  values_to = ".measure",
  location_to = ".location",
  role = "predictor",
  trained = FALSE,
  skip = FALSE,
  id = rand_id("measure_output_long")
)

Arguments

Details

This step is designed convert analytical measurements from their internal data structure to a two column format.

See Also

Other input/output steps: step_measure_input_long(), step_measure_input_wide(), step_measure_output_wide()

Examples

library(dplyr)

data(glucose_bioreactors)
bioreactors_small$batch_sample <- NULL

small_tr <- bioreactors_small[1:200, ]
small_te <- bioreactors_small[201:210, ]

small_rec <-
  recipe(glucose ~ ., data = small_tr) %>%
  update_role(batch_id, day, new_role = "id columns") %>%
  step_measure_input_wide(`400`:`3050`) %>%
  prep()

# Before reformatting:

small_rec %>% bake(new_data = small_te)

# After reformatting:

output_rec <-
  small_rec %>%
  step_measure_output_long() %>%
  prep()

output_rec %>% bake(new_data = small_te)

Description

step_measure_output_wide creates a specification of a recipe step that converts measures to multiple columns (i.e., "wide" format).

Usage

step_measure_output_wide(
  recipe,
  prefix = "measure_",
  role = "predictor",
  trained = FALSE,
  skip = FALSE,
  id = rand_id("measure_output_wide")
)

Arguments

Details

This step is designed convert analytical measurements from their internal data structure to separate columns.

Wide outputs can be helpful when you want to use standard recipes steps with the measuresments, such as recipes::step_pca(), recipes::step_pls(), and so on.

See Also

Other input/output steps: step_measure_input_long(), step_measure_input_wide(), step_measure_output_long()

Examples

library(dplyr)

data(glucose_bioreactors)
bioreactors_small$batch_sample <- NULL

small_tr <- bioreactors_small[1:200, ]
small_te <- bioreactors_small[201:210, ]

small_rec <-
  recipe(glucose ~ ., data = small_tr) %>%
  update_role(batch_id, day, new_role = "id columns") %>%
  step_measure_input_wide(`400`:`3050`) %>%
  prep()

# Before reformatting:

small_rec %>% bake(new_data = small_te)

# After reformatting:

output_rec <-
  small_rec %>%
  step_measure_output_wide() %>%
  prep()

output_rec %>% bake(new_data = small_te)

Description

step_measure_savitzky_golay creates a specification of a recipe step that smooths and filters the measurement sequence.

Usage

step_measure_savitzky_golay(
  recipe,
  role = NA,
  trained = FALSE,
  degree = 3,
  window_side = 11,
  differentiation_order = 0,
  skip = FALSE,
  id = rand_id("measure_savitzky_golay")
)

Arguments

Details

This method can both smooth out random noise and reduce between-predictor correlation. It fits a polynomial to a window of measurements and this results in fewer measurements than the input. Measurements are assumed to be equally spaced.

The polynomial degree should be less than the window size. Also, window size must be greater than polynomial degree. If either case is true, the original argument values are increased to satisfy these conditions (with a warning).

No selectors should be supplied to this step function. The data should be in a special internal format produced by step_measure_input_wide() or step_measure_input_long().

The measurement locations are reset to integer indices starting at one.

Value

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

Tidying

When you tidy() this step, a tibble with columns is returned.

Examples

if (rlang::is_installed("prospectr")) {
  rec <-
    recipe(water + fat + protein ~ ., data = meats_long) %>%
    update_role(id, new_role = "id") %>%
    step_measure_input_long(transmittance, location = vars(channel)) %>%
    step_measure_savitzky_golay(
      differentiation_order = 1,
      degree = 3,
      window_side = 5
    ) %>%
    prep()
}

Description

Subtract baseline using robust fitting method

Usage

subtract_rf_baseline(data, yvar, span = 2/3, maxit = c(5, 5))

Arguments

Value

A dataframe matching column in data plus raw and baseline columns

Examples

meats_long %>% subtract_rf_baseline(yvar = transmittance)

Description

Tidiers for measure steps

Usage

## S3 method for class 'step_measure_input_long'
tidy(x, ...)

## S3 method for class 'step_measure_input_wide'
tidy(x, ...)

## S3 method for class 'step_measure_output_long'
tidy(x, ...)

## S3 method for class 'step_measure_output_wide'
tidy(x, ...)

## S3 method for class 'step_measure_savitzky_golay'
tidy(x, ...)

Arguments

Description

window_side() and differentiation_order() are used with Savitzky-Golay processing.

Usage

window_side(range = c(1L, 5L), trans = NULL)

differentiation_order(range = c(0L, 4L), trans = NULL)

Arguments

Details

This parameter is often used to correct for zero-count data in tables or proportions.

Value

A function with classes "quant_param" and "param".

Examples

window_side()
differentiation_order()