DSL::English::ClassificationWorkflows (Raku package)

zef install DSL::English::ClassificationWorkflows

zef install https://github.com/antononcube/Raku-DSL-English-ClassificationWorkflows.git

use DSL::English::ClassificationWorkflows;

ToClassificationWorkflowCode('make a logistic regression classifier', 'WL::ClCon');

# ClConMakeClassifier[ "LogisticRegression" ]
``` 

Here is a more complicated pipeline specification used to generate the code
for two WL classification systems:

```perl6
my $command = q:to/END/;
use dfTitanic;
split the data with ratio 0.73;
create a logistic regression classifier;
show precision and recall;
show top confusions, misclassified examples, least certain examples;
show ROC plots;
END

say $_.key, "\n", $_.value, "\n"  for ($_ => ToClassificationWorkflowCode($command, $_ ) for <WL::ClCon WL::System>);

# WL::ClCon
# ClConUnit[ dfTitanic ] \[DoubleLongRightArrow]
# ClConSplitData[ "TrainingFraction" -> 0.73 ] \[DoubleLongRightArrow]
# ClConMakeClassifier[ "LogisticRegression" ] \[DoubleLongRightArrow]
# ClConClassifierMeasurements[ {"Precision", "Recall"} ] \[DoubleLongRightArrow] ClConEchoValue[] \[DoubleLongRightArrow]
# ClConClassifierMeasurements[ {"TopConfusions", "MisclassifiedExamples", "LeastCertainExamples"} ] \[DoubleLongRightArrow] ClConEchoValue[] \[DoubleLongRightArrow]
# ClConROCPlot[]
# 
# WL::System
# data = ClConToNormalClassifierData @ dfTitanic;
# {dataTraining, dataTesting} = TakeDrop[ RandomSample[data], Floor[ 0.73 * Length[data] ] ];
# dataValidation = Automatic;
# clObj = Classify[ dataTraining, Method -> "LogisticRegression", ValidationSet -> dataValidation ];
# Echo @ ClassifierMeasurements[clObj, dataTesting, {"Precision", "Recall"} ];
# Echo @ ClassifierMeasurements[clObj, dataTesting, {"TopConfusions", "MisclassifiedExamples", "LeastCertainExamples"} ];
# Echo @ ROCPlot[]

say $_.key, "\n", $_.value, "\n"  for ($_ => ToClassificationWorkflowCode($command, $_ ) for <Bulgarian English Russian>);

# Bulgarian
# използвай данните: dfTitanic 
# раздели данните с тренировъчна част 0.73 
# тренирай класификатор с метод: logistic regression  
# покажи диаграма с приемателните операционни характеристики (ROC)
# 
# English
# use the data: dfTitanic 
# split data with training part 0.73 
# train classifier with method: logistic regression  
# show Receiver Operating Characteristics (ROC) diagram
# 
# Russian
# использовать данные: dfTitanic 
# разделить данные с тренировочная часть 0.73 
# обучить классификатор методом: logistic regression  
# показать диаграмму рабочих характеристик приемника (RХП)

> ToClassificationWorkflowCode --help
# Usage:
#   ToClassificationWorkflowCode <command> [--target=<Str>] [--language=<Str>] [--format=<Str>] -- Translates natural language commands into (machine learning) classification workflow programming code.
#   ToClassificationWorkflowCode <target> <command> [--language=<Str>] [--format=<Str>] -- Both target and command as arguments.
#   
#     <command>           A string with one or many commands (separated by ';').
#     --target=<Str>      Target (programming language with optional library spec.) [default: 'WL-ClCon']
#     --language=<Str>    The natural language to translate from. [default: 'English']
#     --format=<Str>      The format of the output, one of 'automatic', 'code', 'hash', or 'raku'. [default: 'automatic']
#     <target>            Programming language.