| 1234567891011121314151617181920212223242526272829303132333435363738394041424344 |
- """Data preparation, logistic fitting, prediction, and goodness-of-fit."""
- from pathlib import Path
- import numpy as np
- from scipy.io import loadmat
- from ..data import get_data
- from .core import (
- _sigmoid_stable,
- model_p,
- design_matrix,
- nll,
- llf,
- grad_nll,
- hess_nll,
- covariance,
- standard_errors,
- fit_newton,
- goodness_of_fit,
- fit_pack,
- trim_nc_by_value,
- fit_logistic_x,
- predict_curve_x,
- plot_overlay_two_panels_final,
- )
- def load_dataset(project_root=None, percentile=95):
- """Load the thesis SUV predictor and binary outcome arrays."""
- root = Path(project_root) if project_root else Path(__file__).resolve().parents[1]
- suv = loadmat(root / "suv_percentilesSLOthenUWM.mat")
- flags = loadmat(root / "flags_combined.mat")
- x, y = get_data(percentile, suv, flags)
- return np.asarray(x, float).ravel(), np.asarray(y, int).ravel()
- def prepare_full_trim(project_root=None, percentile=95, target=2.48122597, tol=0.05):
- """Return aligned FULL and TRIM datasets."""
- x_full, y_full = load_dataset(project_root, percentile)
- x_trim, y_trim = trim_nc_by_value(x_full, y_full, target=target, tol=tol)
- return {
- "FULL": (x_full, y_full),
- "TRIM": (x_trim, y_trim),
- }
|
