fastlowess-cpp
High-performance LOWESS smoothing
Loading...
Searching...
No Matches
fastlowess::StreamingOptions Struct Reference

Options for streaming LOWESS. More...

#include <fastlowess.hpp>

Inheritance diagram for fastlowess::StreamingOptions:
fastlowess::LowessOptions

Public Attributes

int chunk_size = detail::k_default_chunk_size
 
int overlap = detail::k_default_overlap
 Negative (the default) means "use the library default" (chunk_size / 10, clamped to [1, chunk_size - 10]).
 
std::string merge_strategy
 weighted_average, average, take_first, take_last
 
- Public Attributes inherited from fastlowess::LowessOptions
double fraction = detail::k_default_fraction
 Smoothing fraction (0, 1].
 
int iterations = 3
 Robustness iterations.
 
double delta = NAN
 Interpolation threshold (NaN = auto)
 
std::string weight_function = "tricube"
 
std::string robustness_method = "bisquare"
 
std::string scaling_method = "mad"
 
std::string boundary_policy = "extend"
 
std::string zero_weight_fallback = "use_local_mean"
 
IntervalsOptions intervals
 
std::optional< uint64_t > seed
 
double auto_converge = NAN
 Auto-convergence threshold.
 
std::vector< std::string > outputs
 Optional result components: diagnostics, residuals, weights, derivative, se, and sorted.
 
bool parallel = true
 
std::string backend = "cpu"
 Execution backend: "cpu" (default) or "gpu" (requires the library to be built with the gpu Cargo feature and a Vulkan/Metal/DX12-capable GPU).
 
std::string missing = "error"
 Policy for non-finite (NaN/Inf) values in input data: "error" (default) or "drop".
 
CVOptions cv
 Cross-validation configuration. Empty fractions disable CV.
 
std::vector< double > custom_weights
 Per-observation case weights.
 
bool retain_model = false
 Retain the fitted model's training data, enabling LowessResult::predict_model().
 

Detailed Description

Options for streaming LOWESS.

Member Data Documentation

◆ chunk_size

int fastlowess::StreamingOptions::chunk_size = detail::k_default_chunk_size

◆ merge_strategy

std::string fastlowess::StreamingOptions::merge_strategy
Initial value:
=
"weighted_average"

weighted_average, average, take_first, take_last

◆ overlap

int fastlowess::StreamingOptions::overlap = detail::k_default_overlap

Negative (the default) means "use the library default" (chunk_size / 10, clamped to [1, chunk_size - 10]).


The documentation for this struct was generated from the following file: