|
fastlowess-cpp
High-performance LOWESS smoothing
|
Options for online LOWESS. More...
#include <fastlowess.hpp>
Public Attributes | |
| double | fraction = detail::k_default_fraction |
| Smoothing fraction (0, 1]. | |
| int | iterations = 0 |
| Robustness iterations (requires update_mode = "full") | |
| 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" |
| double | auto_converge = NAN |
| Auto-convergence threshold. | |
| std::vector< std::string > | outputs |
| Optional result components: "weights", "derivative", and/or "se". | |
| IntervalsOptions | intervals |
| std::optional< uint64_t > | seed |
| std::string | missing = "error" |
Policy for non-finite (NaN/Inf) x/y values passed to add_point(): "error" (default) or "drop". | |
| int | window_capacity = detail::k_default_window_capacity |
| int | min_points = detail::k_default_min_points |
| std::string | update_mode = "incremental" |
Options for online LOWESS.
Online LOWESS processes one point at a time, so it has no parallel or backend option, and diagnostics/residuals are always computed for free. outputs = {"se"} or intervals.confidence/intervals.prediction require update_mode = "full".
| std::string fastlowess::OnlineOptions::boundary_policy = "extend" |
| double fastlowess::OnlineOptions::fraction = detail::k_default_fraction |
Smoothing fraction (0, 1].
| IntervalsOptions fastlowess::OnlineOptions::intervals |
| int fastlowess::OnlineOptions::iterations = 0 |
Robustness iterations (requires update_mode = "full")
| int fastlowess::OnlineOptions::min_points = detail::k_default_min_points |
| std::string fastlowess::OnlineOptions::missing = "error" |
Policy for non-finite (NaN/Inf) x/y values passed to add_point(): "error" (default) or "drop".
| std::vector<std::string> fastlowess::OnlineOptions::outputs |
Optional result components: "weights", "derivative", and/or "se".
"se" requires update_mode = "full".
| std::string fastlowess::OnlineOptions::robustness_method = "bisquare" |
| std::string fastlowess::OnlineOptions::scaling_method = "mad" |
| std::optional<uint64_t> fastlowess::OnlineOptions::seed |
| std::string fastlowess::OnlineOptions::update_mode = "incremental" |
| std::string fastlowess::OnlineOptions::weight_function = "tricube" |
| int fastlowess::OnlineOptions::window_capacity = detail::k_default_window_capacity |
| std::string fastlowess::OnlineOptions::zero_weight_fallback = "use_local_mean" |