batch_gd.c File Reference

Demonstrates batch gradient descent for linear regression using ml-framework. More...

#include <stdio.h>
#include <dmlfw.h>
#include <stdlib.h>
#include <unistd.h>

Include dependency graph for batch_gd.c:

#define	FREQUENCY_OF_PRINTING_COST   50000
dmlfw_gradient_descent_options *	get_gradient_descent_options ()
	Creates and configures gradient descent options object.
FILE *	gnuplot
#define	LEARNING_RATE   0.0001
void	load_dataset (dmlfw_mat_double **x, dmlfw_column_vec_double **y)
	Loads the dataset into X (features matrix) and Y (target vector). Adds bias column filled with 1.0.
int	main ()
	Main function to execute batch gradient descent linear regression example.
#define	MODEL_FILE_NAME   "example-1-model.csv"
#define	NUMBER_OF_ITERATIONS   3000000
int	on_iteration_complete (uint64_t iteration_number, void *y, void *predicted_y, void *model, double regularization_parameter)
	Progress callback called on each gradient descent iteration. Logs cost, updates plot files, shows graph via gnuplot at intervals.
void	print_error_and_exit ()
	Prints error string from ml-framework and exits program.
#define	REGULARIZATION_PARAMETER   0.5
#define	SHOW_GRAPH   1
#define	TRAINING_DATASET   "IceCreamSales_training_examples.csv"

Detailed Description

Demonstrates batch gradient descent for linear regression using ml-framework.

Author

Mohammed Daniyal

Version

1.0

Date

2025-09-26

This example loads the IceCreamSales training dataset, trains a linear regression model using batch gradient descent with regularization, tracks cost progress, optionally plots cost and fitted line using gnuplot, and saves the final model to CSV.

Defines custom progress callback for iteration logging and visualization.

Usage: ./batch_gd

Macro Definition Documentation

FREQUENCY_OF_PRINTING_COST

#define FREQUENCY_OF_PRINTING_COST   50000

Frequency of printing/logging cost during training

LEARNING_RATE

#define LEARNING_RATE   0.0001

Learning rate for gradient descent

MODEL_FILE_NAME

#define MODEL_FILE_NAME   "example-1-model.csv"

Output model CSV file path

NUMBER_OF_ITERATIONS

#define NUMBER_OF_ITERATIONS   3000000

Maximum number of iterations for gradient descent

REGULARIZATION_PARAMETER

#define REGULARIZATION_PARAMETER   0.5

Regularization parameter (lambda)

SHOW_GRAPH

#define SHOW_GRAPH   1

Flag to enable (1) or disable (0) graph plotting

TRAINING_DATASET

#define TRAINING_DATASET   "IceCreamSales_training_examples.csv"

Training dataset CSV file path

Function Documentation

get_gradient_descent_options()

dmlfw_gradient_descent_options * get_gradient_descent_options

(

)

Creates and configures gradient descent options object.

Sets learning rate, number of iterations, gradient descent type, and associates the progress callback.

Returns

Pointer to configured dmlfw_gradient_descent_options or NULL on error.

load_dataset()

void load_dataset	(	dmlfw_mat_double **	x,
		dmlfw_column_vec_double **	y
	)

Loads the dataset into X (features matrix) and Y (target vector). Adds bias column filled with 1.0.

Parameters

[out]	x	Pointer to the features matrix pointer.
[out]	y	Pointer to the target column vector pointer.

main()

int main

(

)

Main function to execute batch gradient descent linear regression example.

Loads dataset, creates gradient descent options, fits model, saves model parameters to CSV, and manages resources.

Returns

0 on success.

Examples

example_dmlfw_data_encoder.c, example_dmlfw_error.c, example_dmlfw_linear_regression.c, example_dmlfw_list_double.c, example_dmlfw_mat_double.c, example_dmlfw_math_operations.c, example_dmlfw_ml_configurations.c, example_dmlfw_model_accuracy_score.c, example_dmlfw_operations_double.c, example_dmlfw_scale_double.c, example_dmlfw_utils_file.c, example_dmlfw_vec_double.c, example_dmlfw_vec_string.c, example_mat_string.c, example_set_string.c, and example_utils_string.c.

on_iteration_complete()

int on_iteration_complete	(	uint64_t	iteration_number,
		void *	y,
		void *	predicted_y,
		void *	model,
		double	regularization_parameter
	)

Progress callback called on each gradient descent iteration. Logs cost, updates plot files, shows graph via gnuplot at intervals.

Parameters

iteration_number	Current iteration number.
y	Actual target vector.
predicted_y	Predicted target vector.
model	Current model parameters.
regularization_parameter	Regularization coefficient lambda.

Returns

0 on success, negative to abort.

print_error_and_exit()

void print_error_and_exit

(

)

Prints error string from ml-framework and exits program.

Variable Documentation

gnuplot

FILE* gnuplot

Gnuplot file pointer used for plotting progress