dmlfw_linear_regression.h File Reference

Gradient descent-based linear regression training and prediction. More...

#include <dmlfw_vector.h>
#include <dmlfw_matrix.h>
#include <dmlfw_algorithms.h>

Include dependency graph for dmlfw_linear_regression.h:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Functions
dmlfw_column_vec_double *	dmlfw_linear_regression_fit_using_batch_gradient_descent (dmlfw_gradient_descent_options *gd_options, dmlfw_mat_double *x, dmlfw_column_vec_double *y, double regularization_parameter, dmlfw_column_vec_double *model)
	Fits linear regression model using batch gradient descent.
dmlfw_column_vec_double *	dmlfw_linear_regression_fit_using_mini_batch_gradient_descent (dmlfw_gradient_descent_options *gd_options, double regularization_parameter, dmlfw_column_vec_double *model)
	Fits linear regression model using mini-batch gradient descent.
dmlfw_column_vec_double *	dmlfw_linear_regression_fit_using_stochastic_gradient_descent (dmlfw_gradient_descent_options *gd_options, double regularization_parameter, dmlfw_column_vec_double *model)
	Fits linear regression model using stochastic gradient descent.
dmlfw_column_vec_double *	dmlfw_linear_regression_predict (dmlfw_mat_double *x, dmlfw_column_vec_double *model)
	Predicts output using trained linear regression model.

Detailed Description

Gradient descent-based linear regression training and prediction.

Version

1.0

Date

2025-09-25

This module provides APIs to fit linear regression models using batch, stochastic, and mini-batch gradient descent methods, including support for regularization and user-defined progress callbacks. It also offers prediction functionality from trained models.

Error Handling:

Functions use centralized error reporting. Call dmlfw_error() after function execution to check for errors. For error details, use dmlfw_get_error_string() and dmlfw_get_debug_string().

Ownership:

Returned vectors representing trained models or predictions must be freed using the vector destroy API to avoid memory leaks.

Function Documentation

dmlfw_linear_regression_fit_using_batch_gradient_descent()

dmlfw_column_vec_double * dmlfw_linear_regression_fit_using_batch_gradient_descent	(	dmlfw_gradient_descent_options *	gd_options,
		dmlfw_mat_double *	x,
		dmlfw_column_vec_double *	y,
		double	regularization_parameter,
		dmlfw_column_vec_double *	model
	)

Fits linear regression model using batch gradient descent.

Parameters

gd_options	[in] Gradient descent options (non-NULL).
x	[in] Feature matrix (non-NULL).
y	[in] Target vector (non-NULL).
regularization_parameter	[in] L2 regularization coefficient.
model	[in,out] Optional initial model vector or NULL.

Returns

Pointer to trained model vector or NULL on error.

See also

dmlfw_linear_regression_predict

Usage example:

char err[512], dbg[512];
dmlfw_column_vec_double *model = dmlfw_linear_regression_fit_using_batch_gradient_descent(
    opts, x, y, 0.1, NULL);
if (dmlfw_error()) {
    dmlfw_get_error_string(err, sizeof(err));
    dmlfw_get_debug_string(dbg, sizeof(dbg));
    printf("Error training model: %s\nDebug info: %s\n", err, dbg);
    return NULL;
}
// Use model...
dmlfw_column_vec_double_destroy(model);

Examples

example_dmlfw_linear_regression.c.

dmlfw_linear_regression_fit_using_mini_batch_gradient_descent()

dmlfw_column_vec_double * dmlfw_linear_regression_fit_using_mini_batch_gradient_descent	(	dmlfw_gradient_descent_options *	gd_options,
		double	regularization_parameter,
		dmlfw_column_vec_double *	model
	)

Fits linear regression model using mini-batch gradient descent.

Parameters

gd_options	[in] Gradient descent options (non-NULL).
regularization_parameter	[in] L2 regularization coefficient.
model	[in,out] Optional initial model vector or NULL.

Returns

Pointer to trained model vector or NULL on error.

dmlfw_linear_regression_fit_using_stochastic_gradient_descent()

dmlfw_column_vec_double * dmlfw_linear_regression_fit_using_stochastic_gradient_descent	(	dmlfw_gradient_descent_options *	gd_options,
		double	regularization_parameter,
		dmlfw_column_vec_double *	model
	)

Fits linear regression model using stochastic gradient descent.

Parameters

gd_options	[in] Gradient descent options (non-NULL).
regularization_parameter	[in] L2 regularization coefficient.
model	[in,out] Optional initial model vector or NULL.

Returns

Pointer to trained model vector or NULL on error.

dmlfw_linear_regression_predict()

dmlfw_column_vec_double * dmlfw_linear_regression_predict	(	dmlfw_mat_double *	x,
		dmlfw_column_vec_double *	model
	)

Predicts output using trained linear regression model.

Parameters

x	[in] Feature matrix (non-NULL).
model	[in] Trained model vector (non-NULL).

Returns

Pointer to prediction vector or NULL on error.

Usage example:

char err[512], dbg[512];
dmlfw_column_vec_double *predictions = dmlfw_linear_regression_predict(x, model);
if (dmlfw_error()) {
    dmlfw_get_error_string(err, sizeof(err));
    dmlfw_get_debug_string(dbg, sizeof(dbg));
    printf("Error predicting output: %s\nDebug info: %s\n", err, dbg);
    return NULL;
}
// Use predictions...
dmlfw_column_vec_double_destroy(predictions);

Examples

example_dmlfw_linear_regression.c.