scipy-yli/tests/test_ordinallogit.py

#   scipy-yli: Helpful SciPy utilities and recipes
#   Copyright © 2022  Lee Yingtong Li (RunasSudo)
#
#   This program is free software: you can redistribute it and/or modify
#   it under the terms of the GNU Affero General Public License as published by
#   the Free Software Foundation, either version 3 of the License, or
#   (at your option) any later version.
#
#   This program is distributed in the hope that it will be useful,
#   but WITHOUT ANY WARRANTY; without even the implied warranty of
#   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
#   GNU Affero General Public License for more details.
#
#   You should have received a copy of the GNU Affero General Public License
#   along with this program.  If not, see <https://www.gnu.org/licenses/>.

from pytest import approx

import pandas as pd

import yli

def test_ordinallogit_ucla():
	"""Compare yli.regress with yli.OrdinalLogit for UCLA example at https://stats.oarc.ucla.edu/r/dae/ordinal-logistic-regression/"""
	
	df = pd.read_stata('tests/data/ucla_ologit.dta')
	
	result = yli.regress(yli.OrdinalLogit, df, 'apply', 'pared + public + gpa', exp=False)
	
	assert result.terms['pared'].beta.point == approx(1.04769, abs=0.0001)
	assert result.terms['public'].beta.point == approx(-0.05879, abs=0.001)
	assert result.terms['gpa'].beta.point == approx(0.61594, abs=0.001)
	
	assert result.terms['(Cutoffs)'].categories['unlikely/somewhat likely'].beta.point == approx(2.20391, abs=0.001)
	assert result.terms['(Cutoffs)'].categories['somewhat likely/very likely'].beta.point == approx(4.29936, abs=0.001)
	
	# Confidence intervals compared with Stata 16
	# . ologit apply pared public gpa
	assert result.terms['(Cutoffs)'].categories['unlikely/somewhat likely'].beta.ci_lower == approx(0.6754621, abs=0.001)
	assert result.terms['(Cutoffs)'].categories['unlikely/somewhat likely'].beta.ci_upper == approx(3.731184, abs=0.001)
	assert result.terms['(Cutoffs)'].categories['somewhat likely/very likely'].beta.ci_lower == approx(2.72234, abs=0.001)
	assert result.terms['(Cutoffs)'].categories['somewhat likely/very likely'].beta.ci_upper == approx(5.875195, abs=0.001)
	
	expected_summary = '''              Ordinal Logistic Regression Results              
===============================================================
Dep. Variable:              apply  |  No. Observations:     400
        Model:       OrdinalLogit  |         Df. Model:       5
       Method: Maximum Likelihood  |     Df. Residuals:     395
         Date:         {0:%Y-%m-%d}  |         Pseudo R²:    0.03
         Time:           {0:%H:%M:%S}  |          LL-Model: -358.51
  Std. Errors:         Non-Robust  |           LL-Null: -370.60
                                   |            p (LR): <0.001*
===============================================================
                                β      (95% CI)         p   
------------------------------------------------------------
                      pared    1.05  (0.53 - 1.57)   <0.001*
                     public   -0.06 (-0.64 - 0.53)    0.84  
                        gpa    0.62  (0.10 - 1.13)    0.02* 
                  (Cutoffs)                                 
   unlikely/somewhat likely    2.20  (0.68 - 3.73)    0.005*
somewhat likely/very likely    4.30  (2.72 - 5.88)   <0.001*
------------------------------------------------------------'''.format(result.fitted_dt)
	
	assert result.summary() == expected_summary

def test_brant_ucla():
	"""Compare RegressionResult.brant with R brant library for UCLA example at https://stats.oarc.ucla.edu/r/dae/ordinal-logistic-regression/"""
	
	df = pd.read_stata('tests/data/ucla_ologit.dta')
	result = yli.regress(yli.OrdinalLogit, df, 'apply', 'pared + public + gpa', exp=False)
	brant_result = result.brant()
	
	# df <- read.dta("https://stats.idre.ucla.edu/stat/data/ologit.dta")
	# m <- polr(apply ~ pared + public + gpa, data=df, Hess=TRUE)
	# brant(m)
	
	assert brant_result.tests['Omnibus'].statistic == approx(4.34, abs=0.01)
	assert brant_result.tests['Omnibus'].dof == 3
	assert brant_result.tests['Omnibus'].pvalue == approx(0.23, abs=0.01)
	
	assert brant_result.tests['pared'].statistic == approx(0.13, abs=0.01)
	assert brant_result.tests['pared'].dof == 1
	assert brant_result.tests['pared'].pvalue == approx(0.72, abs=0.01)
	
	assert brant_result.tests['public'].statistic == approx(3.44, abs=0.01)
	assert brant_result.tests['public'].dof == 1
	assert brant_result.tests['public'].pvalue == approx(0.06, abs=0.01)
	
	assert brant_result.tests['gpa'].statistic == approx(0.18, abs=0.01)
	assert brant_result.tests['gpa'].dof == 1
	assert brant_result.tests['gpa'].pvalue == approx(0.67, abs=0.01)
	
	expected_summary = '''          χ²  df     p   
Omnibus  4.34  3   0.23  
pared    0.13  1   0.72  
public   3.44  1   0.06  
gpa      0.18  1   0.67  '''
	
	assert brant_result.summary() == expected_summary