Functions
C-wrapper to C++ implementation of Dbscan algo

Fits a DBSCAN model on an input feature matrix and outputs the labels. More...

Functions

cumlError_t cumlSpDbscanFit (cumlHandle_t handle, float *input, int n_rows, int n_cols, float eps, int min_pts, int *labels, int *core_sample_indices, size_t max_bytes_per_batch, int verbosity)
 
cumlError_t cumlDpDbscanFit (cumlHandle_t handle, double *input, int n_rows, int n_cols, double eps, int min_pts, int *labels, int *core_sample_indices, size_t max_bytes_per_batch, int verbosity)
 

Detailed Description

Fits a DBSCAN model on an input feature matrix and outputs the labels.

Parameters
[in]handlecuml handle to use across the algorithm
[in]inputrow-major input feature matrix
[in]n_rowsnumber of samples in the input feature matrix
[in]n_colsnumber of features in the input feature matrix
[in]epsthe epsilon value to use for epsilon-neighborhood determination
[in]min_ptsminimum number of points to determine a cluster
[out]labels(size n_rows) output labels array
[out]core_sample_indices(size n_rows) output array containing the indices of each core point. If the number of core points is less than n_rows, the right will be padded with -1. Setting this to NULL will prevent calculating the core sample indices
[in]max_mem_bytesthe maximum number of bytes to be used for each batch of the pairwise distance calculation. This enables the trade off between memory usage and algorithm execution time.
[in]verbositySet a verbosity level (higher values means quieter) Refer to cuml/common/logger.hpp for these levels
Returns
CUML_SUCCESS on success and other corresponding flags upon any failures.

Function Documentation

◆ cumlDpDbscanFit()

cumlError_t cumlDpDbscanFit ( cumlHandle_t  handle,
double *  input,
int  n_rows,
int  n_cols,
double  eps,
int  min_pts,
int *  labels,
int *  core_sample_indices,
size_t  max_bytes_per_batch,
int  verbosity 
)

◆ cumlSpDbscanFit()

cumlError_t cumlSpDbscanFit ( cumlHandle_t  handle,
float *  input,
int  n_rows,
int  n_cols,
float  eps,
int  min_pts,
int *  labels,
int *  core_sample_indices,
size_t  max_bytes_per_batch,
int  verbosity 
)