feat: add support for none to indicate no regularization

PR-URL: #13672
Co-authored-by: Athan Reines <kgryte@gmail.com>
Reviewed-by: Athan Reines <kgryte@gmail.com> 
Signed-off-by: Athan Reines <kgryte@gmail.com>
Ref: #12875

Author: nakul-krishnakumar

lib/node_modules/@stdlib/ml/base/sgd-classification/penalties/README.md

@@ -46,14 +46,15 @@ Returns a list of SGD classification penalties.
 
 ```javascript
 var out = penalties();
-// e.g., returns [ 'elasticnet', 'l1', 'l2' ]
+// e.g., returns [ 'elasticnet', 'l1', 'l2', 'none' ]
 ```
 
 The output array contains the following penalties:
 
 -   `elasticnet`: regularization method that linearly combines the L1 and L2 penalties of the lasso and ridge methods.
 -   `l1`: L1 regularization (also called LASSO) leads to sparse models by adding a penalty based on the absolute value of coefficients.
 -   `l2`: L2 regularization (also called ridge regression) encourages smaller, more evenly distributed weights by adding a penalty based on the square of the coefficients.
+-   `none`: no regularization.
 
 </section>
 
@@ -128,6 +129,7 @@ An enumeration of SGD classification penalties with the following fields:
 -   **STDLIB_ML_SGD_CLASSIFICATION_ELASTICNET**: regularization method that linearly combines the L1 and L2 penalties of the lasso and ridge methods.
 -   **STDLIB_ML_SGD_CLASSIFICATION_L1**: L1 regularization (also called LASSO) leads to sparse models by adding a penalty based on the absolute value of coefficients.
 -   **STDLIB_ML_SGD_CLASSIFICATION_L2**: L2 regularization (also called ridge regression) encourages smaller, more evenly distributed weights by adding a penalty based on the square of the coefficients.
+-   **STDLIB_ML_SGD_CLASSIFICATION_NONE**: no regularization.
 
 ```c
 #include "stdlib/ml/base/sgd-classification/penalties.h"

lib/node_modules/@stdlib/ml/base/sgd-classification/penalties/docs/repl.txt

@@ -11,6 +11,7 @@
     - l2: L2 regularization (also called ridge regression) encourages smaller,
     more evenly distributed weights by adding a penalty based on the square of
     the coefficients.
+    - none: no regularization.
 
     Returns
     -------

lib/node_modules/@stdlib/ml/base/sgd-classification/penalties/docs/types/index.d.ts

@@ -25,7 +25,7 @@
 *
 * @example
 * var list = penalties();
-* // e.g., returns [ 'elasticnet', 'l1', 'l2' ]
+* // e.g., returns [ 'elasticnet', 'l1', 'l2', 'none' ]
 */
 declare function penalties(): Array<string>;
 

lib/node_modules/@stdlib/ml/base/sgd-classification/penalties/include/stdlib/ml/base/sgd-classification/penalties.h

@@ -30,7 +30,10 @@ enum STDLIB_ML_SGD_CLASSIFICATION_PENALTY {
 	STDLIB_ML_SGD_CLASSIFICATION_L1,
 
 	// L2 regularization (also called ridge regression) encourages smaller, more evenly distributed weights by adding a penalty based on the square of the coefficients:
-	STDLIB_ML_SGD_CLASSIFICATION_L2
+	STDLIB_ML_SGD_CLASSIFICATION_L2,
+
+	// No regularization:
+	STDLIB_ML_SGD_CLASSIFICATION_NONE
 };
 
 #endif // !STDLIB_ML_BASE_SGD_CLASSIFICATION_PENALTIES_H

lib/node_modules/@stdlib/ml/base/sgd-classification/penalties/lib/data.json

@@ -1,5 +1,6 @@
 [
   "elasticnet",
   "l1",
-  "l2"
+  "l2",
+  "none"
 ]

lib/node_modules/@stdlib/ml/base/sgd-classification/penalties/lib/enum.js

@@ -44,7 +44,10 @@ function enumerated() {
 		'l1': 1,
 
 		// L2 regularization (also called ridge regression) encourages smaller, more evenly distributed weights by adding a penalty based on the square of the coefficients:
-		'l2': 2
+		'l2': 2,
+
+		// No regularization:
+		'none': 3
 	};
 }
 

lib/node_modules/@stdlib/ml/base/sgd-classification/penalties/lib/index.js

@@ -27,7 +27,7 @@
 * var penalties = require( '@stdlib/ml/base/sgd-classification/penalties' );
 *
 * var list = penalties();
-* // e.g., returns [ 'elasticnet', 'l1', 'l2' ]
+* // e.g., returns [ 'elasticnet', 'l1', 'l2', 'none' ]
 */
 
 // MODULES //

lib/node_modules/@stdlib/ml/base/sgd-classification/penalties/lib/main.js

@@ -32,7 +32,7 @@ var DATA = require( './data.json' );
 *
 * @example
 * var list = penalties();
-* // e.g., returns [ 'elasticnet', 'l1', 'l2' ]
+* // e.g., returns [ 'elasticnet', 'l1', 'l2', 'none' ]
 */
 function penalties() {
 	return DATA.slice();

lib/node_modules/@stdlib/ml/base/sgd-classification/penalties/test/test.js

@@ -41,7 +41,8 @@ tape( 'the function returns a list of SGD classification penalties', function te
 	expected = [
 		'elasticnet',
 		'l1',
-		'l2'
+		'l2',
+		'none'
 	];
 	actual = penalties();
 
@@ -64,7 +65,8 @@ tape( 'attached to the main function is an `enum` method to return an object map
 	o = [
 		'elasticnet',
 		'l1',
-		'l2'
+		'l2',
+		'none'
 	];
 	for ( i = 0; i < o.length; i++ ) {
 		t.strictEqual( hasOwnProp( obj, o[ i ] ), true, 'has property `' + o[ i ] + '`' );

lib/node_modules/@stdlib/ml/base/sgd-classification/penalty-enum2str/benchmark/benchmark.js

@@ -37,7 +37,8 @@ bench( pkg, function benchmark( b ) {
 	values = [
 		str2enum( 'elasticnet' ),
 		str2enum( 'l1' ),
-		str2enum( 'l2' )
+		str2enum( 'l2' ),
+		str2enum( 'none' )
 	];
 
 	b.tic();