(experimental) Compute cross-validated correlations for different sample-sizes of a data set. The cross-validation process can be repeated several times to enhance the reliability of the evaluation.

textTrainN(
  x = word_embeddings_4$texts$harmonytext,
  y = Language_based_assessment_data_8$hilstotal,
  sample_percents = c(25, 50, 75, 100),
  n_cross_val = 1,
  seed = 2023
)

Arguments

x

Word embeddings from textEmbed (or textEmbedLayerAggregation). If several word embedding are provided in a list they will be concatenated.

y

Numeric variable to predict.

sample_percents

(numeric) Numeric vector that specifies the percentages of the total number of data points to include in each sample (default = c(25,50,75,100), i.e., correlations are evaluated for 25 the datapoints). The datapoints in each sample are chosen randomly for each new sample.

n_cross_val

(numeric) Value that determines the number of times to repeat the cross-validation. (default = 1, i.e., cross-validation is only performed once). Warning: The training process gets proportionately slower to the number of cross-validations, resulting in a time complexity that increases with a factor of n (n cross-validations).

seed

(numeric) Set different seed (default = 2023).

Value

A tibble containing correlations for each sample. If n_cross_val > 1, correlations for each new cross-validation, along with standard-deviation and mean correlation is included in the tibble. The information in the tibble is visualised via the textTrainNPlot function.

See also

Examples

# Compute correlations for 25%, 50%, 75% and 100% of the data in word_embeddings and perform 
# cross-validation thrice. 

if (FALSE) {
tibble_to_plot <- textTrainN(
      x = word_embeddings_4$texts$harmonytext,
      y = Language_based_assessment_data_8$hilstotal,
      sample_percents = c(25,50,75,100),
      n_cross_val = 3,
)

# tibble_to_plot contains correlation-coefficients for each cross_validation and 
# standard deviation and mean value for each sample. The tibble can be plotted 
# using the testTrainNPlot function.

# Examine tibble
tibble_to_plot
}