Merge pull request #132 from Jacobluke-/master

Proportional Style Plots for DABEST-Python

Author: sangyu

.gitignore

@@ -130,3 +130,5 @@ TEST.ipynb
 TrhCsCh.csv
 Untitled.ipynb
 tt.py
+# Visual Studio
+.vscode/

dabest/__init__.py

@@ -23,4 +23,4 @@
 from ._stats_tools import effsize as effsize
 from ._classes import TwoGroupsEffectSize, PermutationTest
 
-__version__ = "0.3.9999"
+__version__ = "2023.02.14"

dabest/_api.py

@@ -87,6 +87,16 @@ def load(data, idx=None, x=None, y=None, paired=None, id_col=None,
 
     >>> my_data = dabest.load(df, idx=("Control 1", "Test 1"))
 
+    For proportion plot.
+    >>> np.random.seed(88888)
+    >>> N = 10
+    >>> c1 = np.random.binomial(1, 0.2, size=N)
+    >>> t1 = np.random.binomial(1, 0.5, size=N)
+    >>> df = pd.DataFrame({'Control 1' : c1, 'Test 1': t1})
+    >>> my_data = dabest.load(df, idx=("Control 1", "Test 1"),proportional=True)
+
+
+
     '''
     from ._classes import Dabest
 

dabest/_bootstrap_tools.py

@@ -9,92 +9,116 @@
 
 class bootstrap:
     '''Computes the summary statistic and a bootstrapped confidence interval.
+
     Keywords:
         x1, x2: array-like
             The data in a one-dimensional array form. Only x1 is required.
             If x2 is given, the bootstrapped summary difference between
             the two groups (x2-x1) is computed.
             NaNs are automatically discarded.
+
         paired: boolean, default False
             Whether or not x1 and x2 are paired samples. If 'paired' is None then
             the data will not be treated as paired data in the subsequent calculations. 
             If 'paired' is 'baseline', then in each tuple of x, other groups will be
             paired up with the first group (as control). If 'paired' is 'sequential', 
             then in each tuple of x, each group will be paired up with the previous 
             group (as control).
+            
         statfunction: callable, default np.mean
             The summary statistic called on data.
+
         smoothboot: boolean, default False
             Taken from seaborn.algorithms.bootstrap.
             If True, performs a smoothed bootstrap (draws samples from a kernel
             destiny estimate).
+
         alpha: float, default 0.05
             Denotes the likelihood that the confidence interval produced
             does not include the true summary statistic. When alpha = 0.05,
             a 95% confidence interval is produced.
+
         reps: int, default 5000
             Number of bootstrap iterations to perform.
+
     Returns:
         An `bootstrap` object reporting the summary statistics, percentile CIs,
         bias-corrected and accelerated (BCa) CIs, and the settings used.
+
         summary: float
             The summary statistic.
+
         is_difference: boolean
             Whether or not the summary is the difference between two groups.
             If False, only x1 was supplied.
-        is_paired: boolean
-            Whether or not the difference reported is between 2 paired groups.
+
+        is_paired : string, default None
+            The type of the experiment under which the data are obtained
+
         statistic: callable
             The function used to compute the summary.
+
         reps: int
             The number of bootstrap iterations performed.
+
         stat_array: array.
             A sorted array of values obtained by bootstrapping the input arrays.
+
         ci: float
             The size of the confidence interval reported (in percentage).
+
         pct_ci_low, pct_ci_high: floats
             The upper and lower bounds of the confidence interval as computed
             by taking the percentage bounds.
+
         pct_low_high_indices: array
             An array with the indices in `stat_array` corresponding to the
             percentage confidence interval bounds.
+
         bca_ci_low, bca_ci_high: floats
             The upper and lower bounds of the bias-corrected and accelerated
             (BCa) confidence interval. See Efron 1977.
+
         bca_low_high_indices: array
             An array with the indices in `stat_array` corresponding to the BCa
             confidence interval bounds.
+
         pvalue_1samp_ttest: float
             P-value obtained from scipy.stats.ttest_1samp. If 2 arrays were
             passed (x1 and x2), returns 'NIL'.
             See https://docs.scipy.org/doc/scipy-1.0.0/reference/generated/scipy.stats.ttest_1samp.html
+
         pvalue_2samp_ind_ttest: float
             P-value obtained from scipy.stats.ttest_ind.
-            If a single array was given (x1 only), or if `paired` is True,
+            If a single array was given (x1 only), or if `paired` is not None,
             returns 'NIL'.
             See https://docs.scipy.org/doc/scipy-1.0.0/reference/generated/scipy.stats.ttest_ind.html
+
         pvalue_2samp_related_ttest: float
             P-value obtained from scipy.stats.ttest_rel.
-            If a single array was given (x1 only), or if `paired` is False,
+            If a single array was given (x1 only), or if `paired` is None,
             returns 'NIL'.
             See https://docs.scipy.org/doc/scipy-1.0.0/reference/generated/scipy.stats.ttest_rel.html
+
         pvalue_wilcoxon: float
             P-value obtained from scipy.stats.wilcoxon.
-            If a single array was given (x1 only), or if `paired` is False,
+            If a single array was given (x1 only), or if `paired` is None,
             returns 'NIL'.
             The Wilcoxons signed-rank test is a nonparametric paired test of
             the null hypothesis that the related samples x1 and x2 are from
             the same distribution.
             See https://docs.scipy.org/doc/scipy-1.0.0/reference/scipy.stats.wilcoxon.html
+
         pvalue_mann_whitney: float
             Two-sided p-value obtained from scipy.stats.mannwhitneyu.
             If a single array was given (x1 only), returns 'NIL'.
             The Mann-Whitney U-test is a nonparametric unpaired test of the null
             hypothesis that x1 and x2 are from the same distribution.
             See https://docs.scipy.org/doc/scipy-1.0.0/reference/generated/scipy.stats.mannwhitneyu.html
+
     '''
     def __init__(self, x1, x2=None,
-        paired=False,
+        paired=None,
         statfunction=None,
         smoothboot=False,
         alpha_level=0.05,
@@ -136,7 +160,7 @@ def __init__(self, x1, x2=None,
                 if len(x1) != len(x2):
                     raise ValueError('x1 and x2 are not the same length.')
 
-        if (x2 is None) or (paired is True) :
+        if (x2 is None) or (paired is not None) :
 
             if x2 is None:
                 tx = x1
@@ -146,7 +170,7 @@ def __init__(self, x1, x2=None,
                 ttest_2_paired = 'NIL'
                 wilcoxonresult = 'NIL'
 
-            elif paired is True:
+            elif paired is not None:
                 diff = True
                 tx = x2 - x1
                 ttest_single = 'NIL'
@@ -169,7 +193,7 @@ def __init__(self, x1, x2=None,
             pct_low_high = np.nan_to_num(pct_low_high).astype('int')
 
 
-        elif x2 is not None and paired is False:
+        elif x2 is not None and paired is None:
             diff = True
             x2 = pd.Series(x2).dropna()
             # Generate statarrays for both arrays.
@@ -249,7 +273,7 @@ def __repr__(self):
         else:
             stat = self.statistic
 
-        diff_types = {True: 'paired', False: 'unpaired'}
+        diff_types = {'sequential': 'paired', 'baseline': 'paired', None: 'unpaired'}
         if self.is_difference:
             a = 'The {} {} difference is {}.'.format(diff_types[self.is_paired],
                     stat, self.summary)
@@ -265,6 +289,7 @@ def jackknife_indexes(data):
     From the scikits.bootstrap package.
     Given an array, returns a list of arrays where each array is a set of
     jackknife indexes.
+
     For a given set of data Y, the jackknife sample J[i] is defined as the
     data set Y with the ith data point deleted.
     """
@@ -312,4 +337,4 @@ def bca(data, alphas, statarray, statfunction, ostat, reps):
     nvals = np.round((reps-1)*avals)
     nvals = np.nan_to_num(nvals).astype('int')
 
-    return nvals
+    return nvals