Model Parameters Pipeline

This module provides functions to prepare and run a model parameters pipeline for applying sequential data transformations as defined by the Model Parameters specification developed by Big Life Lab.

Workflow

The typical workflow involves two steps:

1. prepare_model_pipeline(): Load and validate model configuration

2. run_model_pipeline(): Apply transformations to data

Required Files

The pipeline requires the following CSV files:

Model Export

Points to variables and model-steps files (columns: fileType, filePath)

Variables

Lists predictor variables (columns: variable, role)

Model Steps

Defines transformation sequence (columns: step, filePath)

Step Parameter Files

Define parameters for each transformation step

See also

• Model Parameters Specification

• Model Parameters Reference Documentation

Examples

if (FALSE) { # \dontrun{
# Basic usage
mod <- prepare_model_pipeline("path/to/model-export.csv")
mod <- run_model_pipeline(mod, data = "path/to/input-data.csv")

# Access transformed data
transformed_data <- mod$df

# Processing multiple datasets with the same model
mod <- prepare_model_pipeline("path/to/model-export.csv")
for (data_file in data_files) {
  result <- run_model_pipeline(mod, data = data_file)
  # Process result$df
}

# Pass a data frame to run_model_pipeline
input_data <- read.csv("path/to/input-data.csv")
mod <- run_model_pipeline(mod, data = input_data)

# Extract logistic predictions (if model includes logistic-regression step)
predictions <- mod$df[, grep("^logistic_", names(mod$df))]
} # }