hdrstats.hdrstats

dat parsing and statistical analysis.

hdrstats.hdrstats.corr_header(lin=True, spearman=False, pearson=False, pvals=True, rbc=False, rmse=False, rrmse=False, mae=False, rmae=False, msd=False, rmsd=False, **kwargs)

generate headers for corr_calc

hdrstats.hdrstats.rel_error(x, y)

hdrstats.hdrstats.corr_calc(x, y, lin=True, spearman=False, pearson=False, pvals=True, rbc=False, rmse=False, rrmse=False, mae=False, rmae=False, msd=False, rmsd=False, **kwargs)

calculate correlations between pairs of data

hdrstats.hdrstats.qq(x, y)

calculate qq plot data

hdrstats.hdrstats.g_err(a, b, c, sigma, scale=1.0, conf=0.95, zero=False)

apply kernel density estimate from b to c at a

hdrstats.hdrstats.g_ssize(a, b, sigma)

hdrstats.hdrstats.kde_cont(b, resample=None, ksigma=None)

hdrstats.hdrstats.error_cont(b, adif, rdif, scale=1, resample=None, ksigma=None)

calculate continuous moving average of error (adif, rdif) based on the kernel density estimate of b, at b or if resample, at resample

hdrstats.hdrstats.error_lgw(x, adif, rdif, coefs, y, t=0.0)

hdrstats.hdrstats.weighted_quantile(d, q, w=None, t=0.0)

hdrstats.hdrstats.weighted_median(d, w=None, t=0.0)

hdrstats.hdrstats.kernel(d, w=None, mi=None, mx=None, n=1000, t=0.0001, bws=0.5)

prepare a gaussian kernel

bws is a scale factor to the bw_method

gaussian kernel selection by Scott’s rule, see: https://docs.scipy.org/doc/scipy/reference/generated/ scipy.stats.gaussian_kde.html

hdrstats.hdrstats.box_stats(d, wg=None, ci=0.75, ciw=0.95, t=0.0)

hdrstats.hdrstats.bootsamps(samples, d, w=None)

hdrstats.hdrstats.conf_box(x, w=None, ci=0.75, ciw=0.95, nsamp=100, t=0.0)

bootstrap a confidence interval for the mean of a weighted sample

hdrstats.hdrstats.quant_box(x, w=None, ci=0.75, ciw=0.95)

create box and whiskers for weighted data

hdrstats.hdrstats.softmax_c(x, coefs)

hdrstats.hdrstats.train_logit(xt, yt, w)