def test_dimension_unit(self):
obs = mlds.MLDSObject(abspath('test.csv'),
boot=False,
save=False,
dimension_unit='stim')
#obs.saveRcommands()
obs.run()
def test_readobjectresults(self):
obs = mlds.MLDSObject(abspath('test.csv'), boot=True, keepfiles=False)
obs.Rdatafile = abspath('output_test.MLDS')
obs.readobjectresults()
self.compare(obs)
def test_bootstrap(self):
obs = mlds.MLDSObject(abspath('test.csv'), boot=True, save=False)
obs.parallel = False
obs.nsamples = 1000
obs.initcommands()
obs.run()
self.compare(obs)
def test_dimension_unit_Rdatafilename(self):
obs = mlds.MLDSObject(abspath('test.csv'),
boot=False,
save=False,
dimension_unit='stim')
obs.getRdatafilename()
assert (obs.Rdatafile == abspath('test_stim_norm_probit.MLDS'))
def test_readdiags(self):
obs = mlds.MLDSObject(abspath('test.csv'), boot=False, keepfiles=False)
obs.Rdatafile = abspath('output_test.MLDS')
obs.readdiags()
self.assertAlmostEqual(obs.AIC, aic, places=d)
self.assertAlmostEqual(obs.DAF, daf, places=d)
def test_readcsv(self):
obs = mlds.MLDSObject(abspath('test.csv'), boot=True, keepfiles=True)
obs.mldsfile = abspath('output_mldsfile.csv')
obs.returncode = 0
obs.readresults()
self.compare(obs)
def test_Rdatafilename4(self):
obs = mlds.MLDSObject(abspath('test.csv'),
boot=False,
save=False,
standardscale=False)
obs.getRdatafilename(force_refit=True)
assert (obs.Rdatafile == abspath('test_unnorm_probit_refit.MLDS'))
def test_simple_othernamecols2(self):
obs = mlds.MLDSObject(abspath('test_cols.csv'), boot=False, save=False)
obs.colnames = {
'stim': ['stim1', 'stim2', 'stim3'],
'response': 'resp'
}
obs.load()
np.testing.assert_almost_equal(obs.scale, scale, decimal=d)
def test_rundiags_nosave(self, saveresiduals=False):
obs = mlds.MLDSObject(abspath('test.csv'), boot=True, keepfiles=False)
obs.parallel = False
obs.nsamples = 250
obs.run()
obs.rundiagnostics(saveresiduals=saveresiduals)
self.assertAlmostEqual(obs.prob, prob, places=1)
os.remove(obs.Rdatafile)
def test_Rdatafilename3(self):
obs = mlds.MLDSObject(abspath('test.csv'),
boot=False,
save=False,
standardscale=False)
obs.linktype = 'cauchit'
obs.getRdatafilename()
assert (obs.Rdatafile == abspath('test_unnorm_cauchit.MLDS'))
def test_bootstrap_correctedCI(self):
obs = mlds.MLDSObject(abspath('test.csv'), boot=True, save=False)
obs.correctedCI = True
obs.parallel = False
obs.nsamples = 1000
obs.run()
np.testing.assert_almost_equal(obs.ci95, ci95_corrected, decimal=d)
np.testing.assert_almost_equal(obs.sigmaci95,
sigmaci95_corrected,
decimal=d)
res = 0.000015 # interpolation resolution. decrease if needed.
# Standards: at which stimulus levels you want to predict thresholds?
sts = np.array([0.25, 0.5, 0.75])
# d' : thresholds will be derived to the following d' performance
dp = [-2, -1, -0.5, 0.5, 1, 2] # d' around the standard at which we will read out thresholds
# (negative values indicate comparisons below the standard, positive values for
# comparisons above the standard)
# type of confidence interval.
citype = 'BCa' # bias-corrected and accelerated CIs
# citype = 'percentile' # raw percentile CI
# %% Estimating scale calling MLDS
ml = mlds.MLDSObject(name, boot=True, standardscale=False, verbose=False)
ml.nsamples = nsamples
ml.load()
# plotting
plt.figure(figsize=(5, 5))
plt.plot(ml.stim, ml.scale, c='k')
yerr = abs(ml.ci95 - ml.mns)
plt.errorbar(ml.stim, ml.mns, yerr=yerr, fmt="none", c='k')
sns.despine()
plt.xlabel('Stimulus')
plt.ylabel('Perceptual scale')
plt.show()
# runs GoF diagnostics
# ml.rundiagnostics()
def test_simple(self):
obs = mlds.MLDSObject(abspath('test.csv'), boot=False, save=False)
obs.run()
np.testing.assert_almost_equal(obs.scale, scale, decimal=d)
def test_Rdatafilename(self):
obs = mlds.MLDSObject(abspath('test.csv'), boot=False, save=False)
print(obs.rootname, obs.filename)
obs.getRdatafilename()
assert (obs.Rdatafile == abspath('test_norm_probit.MLDS'))
# -*- coding: utf-8 -*-
"""
Example use of MLDS class
data.csv was created by simulation.
"""
import matplotlib.pyplot as plt
import mlds
## creating MLDS object and running analysis
obs = mlds.MLDSObject('data.csv', standardscale=False, boot=True, verbose=True)
obs.nsamples = 1000
print("Running analysis...")
# obs.run() # always runs the analysis, overwrites the .MLDS file if exists.
obs.load() # loads results from .MLDS file; runs the analysis if doesn't exist
## plotting
print("plotting...")
plt.figure()
plt.errorbar(obs.stim, obs.mns, yerr=abs(obs.ci95 - obs.mns), color='#4C72B0',
linewidth=2, elinewidth=2)
plt.xlabel('Stimulus')
plt.ylabel('Difference Scale')
plt.xlim((-0.1, 1.1))
plt.show()
