def assess_sample_stability(end_cutoff=3):
ip = get_ipython()
rowavg = ip.user_ns['_rowavg']
tab = [['Sample name', 'Distance', 'Slope of autocorrelation function', 'Stability']]
plt.figure()
for sn in sorted(rowavg):
for dist in sorted(rowavg[sn]):
rowavg_rescaled = rowavg[sn][dist] / rowavg[sn][dist].mean()
rowavg_std = rowavg_rescaled.std() * np.ones_like(rowavg_rescaled)
try:
A, B, stat = nonlinear_leastsquares(np.arange(len(rowavg_rescaled)), rowavg_rescaled, rowavg_std,
lambda x, a, b: a * x + b, [0, 0])
problematic = (A.val > A.err * 3)
except TypeError:
A = 'N/A'
problematic = 2
tab.append([sn, '%.2f' % dist, A, ["\u2713", "\u2718\u2718\u2718\u2718\u2718", '\u274e'][problematic]])
plt.errorbar(np.arange(len(rowavg_rescaled)), rowavg_rescaled,
label=sn + ' %.2f mm' % dist) # ,diags_std[1:])
plt.xlabel('Separation in time (FSN units)')
plt.ylabel('Average discrepancy between curves')
plt.legend(loc='best')
tab = ipy_table.IpyTable(tab)
tab.apply_theme('basic')
display(tab)
plt.show()
def fit(self, function, parameters_init, xname='x', yname='y', dyname='dy', **kwargs):
"""Non-linear least-squares fit to the dataset.
Inputs:
-------
`function`: a callable, corresponding to the function to be fitted.
Should have the following signature::
>>> function(x, par1, par2, par3, ...)
where ``par1``, ``par2``, etc. are the values of the parameters
to be fitted.
`parameters_init`: a sequence (tuple or list) of the initial values
for the parameters to be fitted. Their ordering should be the
same as that of the arguments of `function`
other keyword arguments are given to `nonlinear_leastsquares()`
without any modification.
Outputs:
--------
`par1_fit`, `par2_fit`, etc.: the best fitting values of the parameters.
`statdict`: a dictionary with various status data, such as `R2`, `DoF`,
`Chi2_reduced`, `Covariance`, `Correlation_coeffs` etc. For a full
list, see the help of `sastool.misc.easylsq.nlsq_fit()`
`func_value`: the value of the function at the best fitting parameters,
represented as an instance of the same class as this curve.
Notes:
------
The fitting itself is done via sastool.misc.easylsq.nonlinear_leastsquares()
"""
obj = self.sanitize(fieldname=yname).sanitize(fieldname=xname)
if not hasattr(obj, dyname):
dy = np.ones(len(obj), np.double) * np.nan
else:
dy = getattr(obj, dyname)
ret = nonlinear_leastsquares(getattr(obj, xname), getattr(
obj, yname), dy, function, parameters_init, **kwargs)
funcvalue = type(self)(getattr(obj, xname), ret[-1]['func_value'])
return ret + tuple([funcvalue])
