Functions
Training methods

Fits a HoltWinters model. More...

Functions

cumlError_t cumlHoltWintersSp_fit (cumlHandle_t handle, int n, int batch_size, int frequency, int start_periods, cumlHoltWintersSeasonal_t seasonal, float epsilon, float *data, float *level_ptr, float *trend_ptr, float *season_ptr, float *SSE_error_ptr)
 
cumlError_t cumlHoltWintersDp_fit (cumlHandle_t handle, int n, int batch_size, int frequency, int start_periods, cumlHoltWintersSeasonal_t seasonal, double epsilon, double *data, double *level_ptr, double *trend_ptr, double *season_ptr, double *SSE_error_ptr)
 

Detailed Description

Fits a HoltWinters model.

Parameters
[in]handlecuml handle to use across the algorithm
[in]nn_samples in time-series
[in]batch_sizenumber of time-series in X
[in]frequencynumber of periods in a season of the time-series
[in]start_periodsnumber of seasons to be used for seasonal seed values
[in]seasonaltype of seasonal component (ADDITIVE or MULTIPLICATIVE)
[in]epsilonthe error tolerance value for optimization
[in]datadevice pointer to the data to fit on
[out]level_ptrdevice pointer to array which will hold level components
[out]trend_ptrdevice pointer to array which will hold trend components
[out]season_ptrdevice pointer to array which will hold season components
[out]SSE_error_ptrdevice pointer to array which will hold training SSE error
Returns
CUML_SUCCESS on success and other corresponding flags upon any failures.

Function Documentation

◆ cumlHoltWintersDp_fit()

cumlError_t cumlHoltWintersDp_fit ( cumlHandle_t  handle,
int  n,
int  batch_size,
int  frequency,
int  start_periods,
cumlHoltWintersSeasonal_t  seasonal,
double  epsilon,
double *  data,
double *  level_ptr,
double *  trend_ptr,
double *  season_ptr,
double *  SSE_error_ptr 
)

◆ cumlHoltWintersSp_fit()

cumlError_t cumlHoltWintersSp_fit ( cumlHandle_t  handle,
int  n,
int  batch_size,
int  frequency,
int  start_periods,
cumlHoltWintersSeasonal_t  seasonal,
float  epsilon,
float *  data,
float *  level_ptr,
float *  trend_ptr,
float *  season_ptr,
float *  SSE_error_ptr 
)