predict_quality: Predict writing quality

In writeAlizer: Generate Predicted Writing Quality Scores

Predict writing quality

Description

Run the specified model(s) on preprocessed data and return predictions. Apply scoring models to ReaderBench, Coh-Metrix, and/or GAMET files. Holistic writing quality can be generated from ReaderBench (model = 'rb_mod3all') or Coh-Metrix files (model = 'coh_mod3all'). Also, Total Words Written, Words Spelled Correctly, Correct Word Sequences, and Correct Minus Incorrect Word Sequences can be generated from a GAMET file (model = 'gamet_cws1').

Usage

predict_quality(model, data)

Arguments

model

A string telling which scoring model to use. ReaderBench Model 3 keys also accept a '_v2' suffix. The 'example' key is an offline demonstration. Options are: 'rb_mod1', 'rb_mod2', 'rb_mod3narr', 'rb_mod3exp', 'rb_mod3per', or 'rb_mod3all', for ReaderBench files to generate holistic quality, 'coh_mod1', 'coh_mod2', 'coh_mod3narr', 'coh_mod3exp', 'coh_mod3per', or 'coh_mod3all' for Coh-Metrix files to generate holistic quality, and 'gamet_cws1' to generate Total Words Written (TWW), Words Spelled Correctly (WSC), Correct Word Sequences (CWS) and Correct Minus Incorrect Word Sequences (CIWS) scores from a GAMET file.

data

Data frame returned by import_gamet, import_coh, or import_rb.

Details

Models 2 and 3 center and scale features using the data supplied in this call. Changing the scoring group can change a text's score. A single row or features with no variation can produce missing values. Model 1 and GAMET pass the input through to their saved models without this additional scaling.

The 'example' model is for demonstrating the workflow only. Its preprocessing needs no downloads; prediction requires wa_seed_example_models() first. The temporary files created for the example are cleaned up at the end of the \examples{}.

Value

A data.frame with ID and one column per sub-model prediction. If multiple sub-models are used and all predictions are numeric, an aggregate column named pred_<model>_mean is added (except for "gamet_cws1"). Missing component scores are omitted from the mean; an entirely missing row yields NaN. GAMET returns pred_TWW_gamet, pred_WSC_gamet, pred_CWS_mod1a, and pred_CIWS_mod1a. Predictions are not rounded or clipped.

See Also

import_rb, import_coh, import_gamet

Examples

local({
  old <- options(writeAlizer.mock_dir = NULL, writeAlizer.offline = TRUE)
  on.exit(options(old))
  parent <- tempfile("wa-example-")
  wa_seed_example_models(dir = parent)
  on.exit(unlink(parent, recursive = TRUE), add = TRUE)
  coh <- import_coh(system.file("extdata", "sample_coh.csv", package = "writeAlizer"))
  head(predict_quality("example", coh))
})

# Longer, networked demos
## Not run: 
if (!isTRUE(getOption("writeAlizer.offline", FALSE))) {
  rb <- import_rb(system.file("extdata", "sample_rb.csv", package = "writeAlizer"))
  print(head(predict_quality("rb_mod3all", rb)))

  coh <- import_coh(system.file("extdata", "sample_coh.csv", package = "writeAlizer"))
  print(head(predict_quality("coh_mod3all", coh)))

  gam <- import_gamet(system.file("extdata", "sample_gamet.csv", package = "writeAlizer"))
  print(head(predict_quality("gamet_cws1", gam)))
}

## End(Not run)

writeAlizer documentation built on Sept. 17, 2026, 1:08 a.m.