Small refactorings in API and better errors coverage

Author: cyberosa

dabest/_api.py

@@ -128,13 +128,15 @@ def prop_dataset(
         elif not set(group_names) == set(group.keys()):
             # Check if the group_names provided is the same as the keys of the dict
             raise ValueError("group_names must be the same as the keys of the dict.")
+        
         # Check if the values in the dict are numeric
         if not all(
             [isinstance(group[name], (list, tuple, np.ndarray)) for name in group_names]
         ):
             raise ValueError(
                 "group must be a dict of lists, tuples, or numpy ndarrays of numeric types."
             )
+        
         # Check if the values in the dict only have two elements under each parent key
         if not all([len(group[name]) == 2 for name in group_names]):
             raise ValueError("Each parent key should have only two elements.")
@@ -143,6 +145,7 @@ def prop_dataset(
     else:
         if group_names is None:
             raise ValueError("group_names must be provided if group is not a dict.")
+        
         # Check if the length of group is two times of the length of group_names
         if not len(group) == 2 * len(group_names):
             raise ValueError(

dabest/_bootstrap_tools.py

@@ -88,10 +88,10 @@ def __init__(
             # check x2 is not None:
             if x2 is None:
                 raise ValueError("Please specify x2.")
-            else:
-                x2 = pd.Series(x2).dropna()
-                if len(x1) != len(x2):
-                    raise ValueError("x1 and x2 are not the same length.")
+            
+            x2 = pd.Series(x2).dropna()
+            if len(x1) != len(x2):
+                raise ValueError("x1 and x2 are not the same length.")
 
         if (x2 is None) or (paired is not None):
             if x2 is None:
@@ -102,7 +102,6 @@ def __init__(
                 ttest_2_paired = "NIL"
                 wilcoxonresult = "NIL"
 
-            # elif paired is not None:
             else:  # only two options to enter here
                 diff = True
                 tx = x2 - x1
@@ -234,18 +233,18 @@ def jackknife_indexes(data):
     return (np.delete(base, i) for i in base)
 
 
-def bca(data, alphas, statarray, stat_function, ostat, reps):
+def bca(data, alphas, stat_array, stat_function, ostat, reps):
     """
     Subroutine called to calculate the BCa statistics.
     Borrowed heavily from scikits.bootstrap code.
     """
 
     # The bias correction value.
-    z0 = norm.ppf((1.0 * np.sum(statarray < ostat, axis=0)) / reps)
+    z0 = norm.ppf((1.0 * np.sum(stat_array < ostat, axis=0)) / reps)
 
     # Statistics of the jackknife distribution
-    jackindexes = jackknife_indexes(data[0])
-    jstat = [stat_function(*(x[indexes] for x in data)) for indexes in jackindexes]
+    jack_indexes = jackknife_indexes(data[0])
+    jstat = [stat_function(*(x[indexes] for x in data)) for indexes in jack_indexes]
     jmean = np.mean(jstat, axis=0)
 
     # Acceleration value

dabest/_delta_objects.py

@@ -409,8 +409,7 @@ def __init__(self, effectsizedataframe, permutation_count,
         # weights
         self.__bootstraps_weighted_delta = ci2g.calculate_weighted_delta(
                                                           self.__group_var, 
-                                                          self.__bootstraps, 
-                                                          self.__resamples)
+                                                          self.__bootstraps)
 
         # Compute the weighted average mean difference based on the raw data
         self.__difference = es.weighted_delta(self.__effsizedf["difference"],

dabest/_effsize_objects.py

@@ -663,10 +663,12 @@ def permutations(self):
 
     @property
     def permutations_var(self):
+         # TODO Missing docstring
         return self.__PermutationTest_result.permutations_var
 
     @property
     def proportional_difference(self):
+         # TODO Missing docstring
         try:
             return self.__proportional_difference
         except AttributeError:

dabest/_stats_tools/confint_1group.py

@@ -38,6 +38,7 @@ def compute_1group_jackknife(x, func, *args, **kwargs):
 
 
 def compute_1group_acceleration(jack_dist):
+    # TODO is it needed a function to just call one line?
     from . import confint_2group_diff as ci_2g
 
     return ci_2g._calc_accel(jack_dist)

dabest/_stats_tools/confint_2group_diff.py

@@ -98,6 +98,7 @@ def compute_meandiff_jackknife(x0, x1, is_paired, effect_size):
 
 
 def _calc_accel(jack_dist):
+    # TODO Missing docstring
     jack_mean = npmean(jack_dist)
 
     numer = npsum((jack_mean - jack_dist) ** 3)
@@ -283,17 +284,17 @@ def compute_interval_limits(bias, acceleration, n_boots, ci=95):
     if isnan(low) or isnan(high):
         return low, high
 
-    else:
-        low = int(norm.cdf(low) * n_boots)
-        high = int(norm.cdf(high) * n_boots)
-        return low, high
+    
+    low = int(norm.cdf(low) * n_boots)
+    high = int(norm.cdf(high) * n_boots)
+    return low, high
 
 
 def calculate_group_var(control_var, control_N, test_var, test_N):
     return control_var / control_N + test_var / test_N
 
 
-def calculate_weighted_delta(group_var, differences, resamples):
+def calculate_weighted_delta(group_var, differences):
     """
     Compute the weighted deltas.
     """

dabest/_stats_tools/effsize.py

@@ -64,7 +64,7 @@ def two_group_difference(control:list|tuple|np.ndarray, #Accepts lists, tuples,
     if effect_size == "mean_diff":
         return func_difference(control, test, np.mean, is_paired)
 
-    elif effect_size == "median_diff":
+    if effect_size == "median_diff":
         mes1 = "Using median as the statistic in bootstrapping may " + \
                 "result in a biased estimate and cause problems with " + \
                 "BCa confidence intervals. Consider using a different statistic, such as the mean.\n"
@@ -74,21 +74,21 @@ def two_group_difference(control:list|tuple|np.ndarray, #Accepts lists, tuples,
         warnings.warn(message=mes1+mes2, category=UserWarning)
         return func_difference(control, test, np.median, is_paired)
 
-    elif effect_size == "cohens_d":
+    if effect_size == "cohens_d":
         return cohens_d(control, test, is_paired)
 
-    elif effect_size == "cohens_h":
+    if effect_size == "cohens_h":
         return cohens_h(control, test)
 
-    elif effect_size == "hedges_g" or effect_size == "delta_g":
+    if effect_size == "hedges_g" or effect_size == "delta_g":
         return hedges_g(control, test, is_paired)
 
-    elif effect_size == "cliffs_delta":
+    if effect_size == "cliffs_delta":
         if is_paired:
             err1 = "`is_paired` is not None; therefore Cliff's delta is not defined."
             raise ValueError(err1)
-        else:
-            return cliffs_delta(control, test)
+        
+        return cliffs_delta(control, test)
 
 
 # %% ../../nbs/API/effsize.ipynb 6
@@ -129,10 +129,10 @@ def func_difference(control:list|tuple|np.ndarray, # NaNs are automatically disc
 
         return func(test - control)
 
-    else:
-        control = control[~np.isnan(control)]
-        test    = test[~np.isnan(test)]
-        return func(test) - func(control)
+    
+    control = control[~np.isnan(control)]
+    test    = test[~np.isnan(test)]
+    return func(test) - func(control)
 
 
 # %% ../../nbs/API/effsize.ipynb 7
@@ -209,7 +209,8 @@ def cohens_d(control:list|tuple|np.ndarray,
     else:
         M = np.mean(test) - np.mean(control)
         divisor = pooled_sd
-        
+    
+    # TODO what if divisor = 0?
     return M / divisor
 
 # %% ../../nbs/API/effsize.ipynb 8
@@ -289,7 +290,6 @@ def cliffs_delta(control:list|tuple|np.ndarray,
     See [here](https://en.wikipedia.org/wiki/Effect_size#Effect_size_for_ordinal_data)
     """
 
-
     # Convert to numpy arrays for speed.
     # NaNs are automatically dropped.
     if ~isinstance(control, np.ndarray):
@@ -307,7 +307,6 @@ def cliffs_delta(control:list|tuple|np.ndarray,
     U, _ = mannwhitneyu(t, c, alternative='two-sided')
     cliffs_delta = ((2 * U) / (control_n * test_n)) - 1
 
-
     return cliffs_delta
 
 
@@ -332,7 +331,7 @@ def _compute_standardizers(control, test):
     # For paired standardized mean difference.
     average = np.sqrt((control_var + test_var) / 2)
 
-    return pooled, average # indent if you implement above code chunk.
+    return pooled, average 
 
 # %% ../../nbs/API/effsize.ipynb 12
 def _compute_hedges_correction_factor(n1, 

dabest/misc_tools.py

@@ -33,6 +33,7 @@ def unpack_and_add(l, c):
 
 
 def print_greeting():
+    # TODO missing docstring
     from .__init__ import __version__
 
     line1 = "DABEST v{}".format(__version__)