SQL functions for support vector machines. More...

Functions
float8	svm_dot (float8[] x, float8[] y)
	Dot product kernel function. More...

float8	svm_polynomial (float8[] x, float8[] y, float8 degree)
	Polynomial kernel function. More...

float8	svm_gaussian (float8[] x, float8[] y, float8 gamma)
	Gaussian kernel function. More...

void	svm_drop_model (text model_table)
	Drops all tables pertaining to a model. More...

float8	svm_predict (text model_table, float8[] ind)
	Evaluates a support-vector model on a given data point. More...

set< svm_model_pr >	svm_predict_combo (text model_table, float8[] ind)
	Evaluates multiple support-vector models on a data point. More...

set< svm_reg_result >	svm_regression (text input_table, text model_table, bool parallel, text kernel_func)
	This is the support vector regression function. More...

set< svm_reg_result >	svm_regression (text input_table, text model_table, bool parallel, text kernel_func, bool verbose, float8 eta, float8 nu, float8 slambda)
	This is the support vector regression function. More...

set< svm_cls_result >	svm_classification (text input_table, text model_table, bool parallel, text kernel_func)
	This is the support vector classification function. More...

set< svm_cls_result >	svm_classification (text input_table, text model_table, bool parallel, text kernel_func, bool verbose, float8 eta, float8 nu)
	This is the support vector classification function. More...

set< svm_nd_result >	svm_novelty_detection (text input_table, text model_table, bool parallel, text kernel_func)
	This is the support vector novelty detection function. More...

set< svm_nd_result >	svm_novelty_detection (text input_table, text model_table, bool parallel, text kernel_func, bool verbose, float8 eta, float8 nu)
	This is the support vector novelty detection function. More...

text	svm_predict_batch (text input_table, text data_col, text id_col, text model_table, text output_table, bool parallel)
	Scores the data points stored in a table using a learned support-vector model. More...

void	svm_data_normalization (text input_table)
	Normalizes the data stored in a table, and save the normalized data in a new table. More...

set< lsvm_sgd_result >	lsvm_classification (text input_table, text model_table, bool parallel)
	This is the linear support vector classification function. More...

set< lsvm_sgd_result >	lsvm_classification (text input_table, text model_table, bool parallel, bool verbose, float8 eta, float8 reg)
	This is the linear support vector classification function. More...

text	lsvm_predict_batch (text input_table, text data_col, text id_col, text model_table, text output_table, bool parallel)
	Scores the data points stored in a table using a learned linear support-vector model. More...

float8	lsvm_predict (text model_table, float8[] ind)
	Evaluates a linear support-vector model on a given data point. More...

set< svm_model_pr >	lsvm_predict_combo (text model_table, float8[] ind)
	Evaluates multiple linear support-vector models on a data point. More...

Detailed Description

See Also: For an introduction to Support vector machines (SVMs) and related kernel methods, see the module description Support Vector Machines.

Definition in file online_sv.sql_in.

Function Documentation

set<lsvm_sgd_result> lsvm_classification	(	text	input_table,
		text	model_table,
		bool	parallel
	)

Parameters

input_table	The name of the table/view with the training data
model_table	The name of the table under which we want to store the learned model
parallel	A flag indicating whether the system should learn multiple models in parallel

Returns: A summary of the learning process

Definition at line 875 of file online_sv.sql_in.

set<lsvm_sgd_result> lsvm_classification	(	text	input_table,
		text	model_table,
		bool	parallel,
		bool	verbose,
		float8	eta,
		float8	reg
	)

Parameters

input_table	The name of the table/view with the training data
model_table	The name of the table under which we want to store the learned model
parallel	A flag indicating whether the system should learn multiple models in parallel
verbose	Verbosity of reporting
eta	Initial learning rate in (0,1]
reg	Regularization parameter, often chosen by cross-validation

Returns: A summary of the learning process

Definition at line 900 of file online_sv.sql_in.

float8 lsvm_predict	(	text	model_table,
		float8[]	ind
	)

Parameters

model_table	The table storing the learned model \( f \) to be used
ind	The data point \( \boldsymbol x \)

Returns: This function returns \( f(\boldsymbol x) \)

Definition at line 947 of file online_sv.sql_in.

text lsvm_predict_batch	(	text	input_table,
		text	data_col,
		text	id_col,
		text	model_table,
		text	output_table,
		bool	parallel
	)

Parameters

input_table	Name of table/view containing the data points to be scored
data_col	Name of column in input_table containing the data points
id_col	Name of column in input_table containing the integer identifier of data points
model_table	Name of table where the learned model to be used is stored
output_table	Name of table to store the results
parallel	A flag indicating whether the model to be used was learned in parallel

Returns: Textual summary of the algorithm run

Definition at line 927 of file online_sv.sql_in.

set<svm_model_pr> lsvm_predict_combo	(	text	model_table,
		float8[]	ind
	)

Parameters

model_table	The table storing the learned models to be used.
ind	The data point \( \boldsymbol x \)

Returns: This function returns a table, a row for each model. Moreover, the last row contains the average value, over all models.

The different models are assumed to be named model_table0, model_table1, ....

Definition at line 968 of file online_sv.sql_in.

set<svm_cls_result> svm_classification	(	text	input_table,
		text	model_table,
		bool	parallel,
		text	kernel_func
	)

Parameters

input_table	The name of the table/view with the training data
model_table	The name of the table under which we want to store the learned model
parallel	A flag indicating whether the system should learn multiple models in parallel
kernel_func	Kernel function

Returns: A summary of the learning process

Definition at line 665 of file online_sv.sql_in.

set<svm_cls_result> svm_classification	(	text	input_table,
		text	model_table,
		bool	parallel,
		text	kernel_func,
		bool	verbose,
		float8	eta,
		float8	nu
	)

Parameters

input_table	The name of the table/view with the training data
model_table	The name of the table under which we want to store the learned model
parallel	A flag indicating whether the system should learn multiple models in parallel
kernel_func	Kernel function
verbose	Verbosity of reporting
eta	Learning rate in (0,1]
nu	Compression parameter in (0,1] associated with the fraction of training data that will become support vectors

Returns: A summary of the learning process

Definition at line 692 of file online_sv.sql_in.

void svm_data_normalization ( text input_table)

Parameters

input_table Name of table/view containing the data points to be scored

Definition at line 855 of file online_sv.sql_in.

float8 svm_dot	(	float8[]	x,
		float8[]	y
	)

Parameters

x	The data point \( \boldsymbol x \)
y	The data point \( \boldsymbol y \)

Returns: Returns dot product of the two data points.

Definition at line 428 of file online_sv.sql_in.

void svm_drop_model ( text model_table)

Parameters

model_table The table to be dropped.

Definition at line 555 of file online_sv.sql_in.

float8 svm_gaussian	(	float8[]	x,
		float8[]	y,
		float8	gamma
	)

Parameters

x	The data point \( \boldsymbol x \)
y	The data point \( \boldsymbol y \)
gamma	The spread \( \gamma \)

Returns: Returns \( K(\boldsymbol x,\boldsymbol y)=exp(-\gamma || \boldsymbol x \cdot \boldsymbol y ||^2 ) \)

Definition at line 452 of file online_sv.sql_in.

set<svm_nd_result> svm_novelty_detection	(	text	input_table,
		text	model_table,
		bool	parallel,
		text	kernel_func
	)

Parameters

input_table	The name of the table/view with the training data
model_table	The name of the table under which we want to store the learned model
parallel	A flag indicating whether the system should learn multiple models in parallel
kernel_func	Kernel function

Returns: A summary of the learning process

Definition at line 716 of file online_sv.sql_in.

set<svm_nd_result> svm_novelty_detection	(	text	input_table,
		text	model_table,
		bool	parallel,
		text	kernel_func,
		bool	verbose,
		float8	eta,
		float8	nu
	)

Parameters

input_table	The name of the table/view with the training data
model_table	The name of the table under which we want to store the learned model
parallel	A flag indicating whether the system should learn multiple models in parallel
kernel_func	Kernel function
verbose	Verbosity of reporting
eta	Learning rate in (0,1]
nu	Compression parameter in (0,1] associated with the fraction of training data that will become support vectors

Returns: A summary of the learning process

Definition at line 743 of file online_sv.sql_in.

float8 svm_polynomial	(	float8[]	x,
		float8[]	y,
		float8	degree
	)

Parameters

x	The data point \( \boldsymbol x \)
y	The data point \( \boldsymbol y \)
degree	The degree \( d \)

Returns: Returns \( K(\boldsymbol x,\boldsymbol y)=(\boldsymbol x \cdot \boldsymbol y)^d \)

Definition at line 440 of file online_sv.sql_in.

float8 svm_predict	(	text	model_table,
		float8[]	ind
	)

Parameters

model_table	The table storing the learned model \( f \) to be used
ind	The data point \( \boldsymbol x \)

Returns: This function returns \( f(\boldsymbol x) \)

Definition at line 569 of file online_sv.sql_in.

text svm_predict_batch	(	text	input_table,
		text	data_col,
		text	id_col,
		text	model_table,
		text	output_table,
		bool	parallel
	)

Parameters

input_table	Name of table/view containing the data points to be scored
data_col	Name of column in input_table containing the data points
id_col	Name of column in input_table containing the integer identifier of data points
model_table	Name of table where the learned model to be used is stored
output_table	Name of table to store the results
parallel	A flag indicating whether the model to be used was learned in parallel

Returns: Textual summary of the algorithm run

Definition at line 770 of file online_sv.sql_in.

set<svm_model_pr> svm_predict_combo	(	text	model_table,
		float8[]	ind
	)

Parameters

model_table	The table storing the learned models to be used.
ind	The data point \( \boldsymbol x \)

Returns: This function returns a table, a row for each model. Moreover, the last row contains the average value, over all models.

The different models are assumed to be named model_table1, model_table2, ....

Definition at line 590 of file online_sv.sql_in.

set<svm_reg_result> svm_regression	(	text	input_table,
		text	model_table,
		bool	parallel,
		text	kernel_func
	)

Parameters

input_table	The name of the table/view with the training data
model_table	The name of the table under which we want to store the learned model
parallel	A flag indicating whether the system should learn multiple models in parallel
kernel_func	Kernel function

Returns: A summary of the learning process

Definition at line 613 of file online_sv.sql_in.

set<svm_reg_result> svm_regression	(	text	input_table,
		text	model_table,
		bool	parallel,
		text	kernel_func,
		bool	verbose,
		float8	eta,
		float8	nu,
		float8	slambda
	)

Parameters

input_table	The name of the table/view with the training data
model_table	The name of the table under which we want to store the learned model
parallel	A flag indicating whether the system should learn multiple models in parallel
kernel_func	Kernel function
verbose	Verbosity of reporting
eta	Learning rate in (0,1]
nu	Compression parameter in (0,1] associated with the fraction of training data that will become support vectors
slambda	Regularisation parameter

Returns: A summary of the learning process

Definition at line 641 of file online_sv.sql_in.

Functions

Detailed Description

Function Documentation