unifiedml in R: A Unified Machine Learning Interface

T. Moudiki

1 week ago

[This article was first published on T. Moudiki's Webpage - R, and kindly contributed to R-bloggers]. (You can report issue about the content on this page here)

Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.

unifiedml

A Unified Machine Learning Interface for R

Overview

unifiedml provides a consistent, sklearn-like interface for virtually any machine learning model in R.

It eliminates the need to remember different function signatures across packages by automatically detecting the appropriate interface (formula vs matrix) and task type (regression vs classification).

Key Features

For now:

Automatic Task Detection: Automatically detects regression vs classification based on response variable type (numeric → regression, factor → classification)
Universal Interface: Works seamlessly with glmnet, randomForest, e1071::svm, and other popular ML packages with formula or matrix interface
Built-in Cross-Validation: Consistent cross_val_score() function with automatic metric selection
Model Interpretability: Numerical derivatives and statistical significance testing via summary()
Partial Dependence Plots: Visualize feature effects with plot()
Method Chaining: Clean, pipeable syntax with R6 classes

devtools::install_github("Techtonique/unifiedml")

install.packages(c("glmnet", "randomForest"))

install.packages("e1071")

library(unifiedml) # this package
library(glmnet)
library(randomForest)
library(e1071)

# ------------------------------------------------------------
# REGRESSION EXAMPLES
# ------------------------------------------------------------

cat("\n=== REGRESSION EXAMPLES ===\n\n")

# Example 1: Synthetic data (numeric y → automatic regression)
set.seed(123)
X <- MASS::Boston[, -ncol(MASS::Boston)]
y <- MASS::Boston$medv

# glmnet regression
cat("1. Ridge Regression (glmnet) - Auto-detected: Regression\n")
mod1 <- Model$new(glmnet::glmnet)  # No task parameter needed!
mod1$fit(X, y, alpha = 0, lambda = 0.1)
mod1$print()
cat("\nPredictions:\n")
print(head(mod1$predict(X)))
cat("\n")
mod1$summary()
mod1$plot(feature = 1)
(cv1 <- cross_val_score(mod1, X, y, cv = 5L))  # Auto-uses RMSE
cat("\nMean RMSE:", mean(cv1), "\n\n")

# randomForest regression
cat("\n2. Random Forest Regression - Auto-detected: Regression\n")
mod2 <- Model$new(randomForest::randomForest)  # No task parameter!
mod2$fit(X, y, ntree = 50)
mod2$print()
cat("\n")
mod2$summary(h = 0.01)

# ------------------------------------------------------------
# CLASSIFICATION EXAMPLES
# ------------------------------------------------------------

cat("\n\n=== CLASSIFICATION EXAMPLES ===\n\n")

# Example: Iris dataset (factor y → automatic classification)
data(iris)

# Binary classification with factor
cat("3. Binary Classification with Factor Response\n")
iris_binary <- iris[iris$Species %in% c("setosa", "versicolor"), ]
X_binary <- as.matrix(iris_binary[, 1:4])
y_binary <- iris_binary$Species  # factor → classification

# Multi-class classification
cat("4. Multi-class Classification\n")
X_multi <- as.matrix(iris[, 1:4])
y_multi <- iris$Species  # factor with 3 levels → multi-class classification

mod4 <- Model$new(randomForest::randomForest)  # No task parameter!
mod4$fit(X_multi, y_multi, ntree = 50)
mod4$print()
(cv4 <- cross_val_score(mod4, X_multi, y_multi, cv = 5L))  # Auto-uses accuracy
cat("\nMean Accuracy:", mean(cv4), "\n")


y_multi_numeric <- as.numeric(y_multi)
mod4 <- Model$new(glmnet::glmnet)  # No task parameter!
mod4$fit(X_multi, y_multi_numeric, family="multinomial")
mod4$print()
(cv4 <- cross_val_score(mod4, X_multi, y_multi_numeric, cv = 5L))  # Auto-uses accuracy
cat("\nMean Accuracy:", mean(cv4), "\n")

=== REGRESSION EXAMPLES ===

1. Ridge Regression (glmnet) - Auto-detected: Regression
Model Object
------------
Model function: self$model_fn 
Fitted: TRUE 
Task: regression 
Training samples: 506 
Features: 13 

Predictions:
       1        2        3        4        5        6 
30.12476 25.01360 30.57030 28.68765 28.04710 25.31151 


Model Summary - Numerical Derivatives
======================================
Task: regression 
Samples: 506 | Features: 13 
Step size (h): 0.01 

 Feature Mean_Derivative    Std_Error       t_value p_value Significance
    crim   -1.032452e-01 7.837905e-15 -1.317256e+13       0          ***
      zn    4.322719e-02 5.216968e-15  8.285883e+12       0          ***
   indus    2.743581e-03 1.273890e-14  2.153704e+11       0          ***
    chas    2.753495e+00 6.050646e-15  4.550745e+14       0          ***
     nox   -1.656232e+01 1.293028e-14 -1.280894e+15       0          ***
      rm    3.868607e+00 9.300418e-15  4.159605e+14       0          ***
     age   -4.129908e-04 1.108270e-14 -3.726444e+10       0          ***
     dis   -1.411492e+00 4.776813e-15 -2.954882e+14       0          ***
     rad    2.655385e-01 7.851465e-15  3.382025e+13       0          ***
     tax   -1.038490e-02 8.323530e-15 -1.247656e+12       0          ***
 ptratio   -9.325559e-01 6.254096e-15 -1.491112e+14       0          ***
   black    9.272792e-03 3.489110e-15  2.657638e+12       0          ***
   lstat   -5.149643e-01 4.667193e-15 -1.103371e+14       0          ***

Significance codes: 0 '***' 0.01 '**' 0.05 '*' 0.1 ' ' 1
  |======================================================================| 100%

Mean RMSE: 5.490208 


2. Random Forest Regression - Auto-detected: Regression
Model Object
------------
Model function: self$model_fn 
Fitted: TRUE 
Task: regression 
Training samples: 506 
Features: 13 


Model Summary - Numerical Derivatives
======================================
Task: regression 
Samples: 506 | Features: 13 
Step size (h): 0.01 

 Feature Mean_Derivative   Std_Error    t_value      p_value Significance
    crim      0.27808959 0.329662424  0.8435587 3.993155e-01             
      zn      0.00000000 0.000000000        NaN          NaN         <NA>
   indus      0.03961792 0.030839187  1.2846616 1.994995e-01            *
    chas      0.00000000 0.000000000        NaN          NaN         <NA>
     nox     -7.59855104 1.939451552 -3.9178865 1.016265e-04          ***
      rm      4.33226614 0.538906676  8.0389914 6.471192e-15          ***
     age     -0.02569829 0.023089867 -1.1129682 2.662516e-01             
     dis     -0.71310569 0.405677973 -1.7578122 7.938528e-02           **
     rad      0.01739130 0.017453435  0.9964402 3.195135e-01             
     tax     -0.01361660 0.008303022 -1.6399573 1.016368e-01            *
 ptratio     -0.03075099 0.024143264 -1.2736881 2.033599e-01             
   black      0.05649539 0.045334095  1.2462009 2.132684e-01             
   lstat     -0.47499012 0.124407440 -3.8180202 1.512437e-04          ***

Significance codes: 0 '***' 0.01 '**' 0.05 '*' 0.1 ' ' 1


=== CLASSIFICATION EXAMPLES ===

3. Binary Classification with Factor Response
4. Multi-class Classification
Model Object
------------
Model function: self$model_fn 
Fitted: TRUE 
Task: classification 
Training samples: 150 
Features: 4 
Classes: setosa, versicolor, virginica 
Class distribution:

    setosa versicolor  virginica 
        50         50         50 
  |======================================================================| 100%

Mean Accuracy: 0.96 
Model Object
------------
Model function: self$model_fn 
Fitted: TRUE 
Task: regression 
Training samples: 150 
Features: 4 
  |======================================================================| 100%

Mean Accuracy: 0.3228121

To leave a comment for the author, please follow the link and comment on their blog: T. Moudiki's Webpage - R.

R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.

Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.

unifiedml

Overview

Key Features

Related