Add files via upload

Author: nbashir97

src/effectsize.py

@@ -0,0 +1,112 @@
+import pandas
+from functions import list_filter, compute_continuous, compute_categorical
+
+#%%
+
+def compute(data,
+            group,
+            continuous = [],
+            categorical = [],
+            skewed = [],
+            weights = None,
+            decimals = 2,
+            intervals = None):
+    
+    """
+    
+    Computes SDs for all specified variables
+    
+    Parameters:
+        data (dataframe): Pandas DataFrame containing observations (rows) and variables (columns)
+        exposure (str): Variable defining exposed and unexposed
+        continuous (list): List of string items which are names of the continuous variables for which the SD should be computed
+        categorical (list): List of string items which are names of the categorical variables for which the SD should be computed
+        skewed (list): List of string items which are names of the continuous variables which have a skewed distribution (ranked SD computed)
+        weights (None or str): Variable defining weights for each observation (otherwise assumed to be equally weighted)
+        decimals (int): Number of decimal places which should be computed
+        intervals (None or float): Whether CIs should be computed and with what coverage e.g. for 95% CI, intervals = 0.95
+    
+    Returns:
+        Returns a Pandas DataFrame containing the computed SDs (plus CIs, if specified)
+
+    """
+      
+    # Asserting input types
+    
+    assert type(data) == pandas.DataFrame or type(data) == pandas.core.frame.DataFrame, "Data must be specified as a Pandas DataFrame"        
+    assert type(group) == str, "Group variable must be specified as a string"
+    assert type(continuous) == list and type(categorical) == list and type(skewed) == list, "Variable names must be specified inside lists"
+    assert weights == None or type(weights) == str, "If weight variable is present, it must be specified as a string"
+    assert type(decimals) == int, "Number of decimal places must be specified as an integer"
+    assert intervals == None or (intervals > 0 and intervals < 1), "CIs must be specified as None or in range (0,1) e.g. for 95% CI, intervals = 0.95"
+        
+    # Get combined list of variables and sort them into the order in which they appear in the dataframe
+    
+    specified_variables = (continuous + categorical).copy()
+    all_variables = list(data)
+    
+    for variable in specified_variables:
+        assert type(variable) == str, "The variable names inside lists must all be specified as strings"
+        if variable not in all_variables: 
+            print("The following variable was not computed as it could not be found in dataframe columns:", variable)        
+    
+    ordered_variables = list_filter(list1 = all_variables, list2 = specified_variables)
+    
+    # Computing the standardized difference
+    
+    results = []
+    
+    for variable in ordered_variables:
+
+        if variable in continuous:
+            
+            if variable in skewed:
+                
+                stdiff = compute_continuous(data = data,
+                                            group = group,
+                                            variable = variable,
+                                            skewed = True,
+                                            weights = weights,
+                                            decimals = decimals,
+                                            intervals = intervals)
+                
+                results.append(stdiff)
+                
+            else:
+                
+                stdiff = compute_continuous(data = data,
+                                            group = group,
+                                            variable = variable,
+                                            skewed = False,
+                                            weights = weights,
+                                            decimals = decimals,
+                                            intervals = intervals)
+                
+                results.append(stdiff)
+        
+        else:
+            
+            stdiff = compute_categorical(data = data,
+                                         group = group, 
+                                         variable = variable,
+                                         weights = weights,
+                                         decimals = decimals,
+                                         intervals = intervals)
+            
+            results.append(stdiff)
+    
+    results = pandas.DataFrame(data = results)
+    results.set_axis([ordered_variables], axis = 0, inplace = True)
+    
+    # Computing the CIs
+    
+    if intervals == None:
+        
+        results.set_axis(['ES'], axis = 1, inplace = True)
+    
+    else:
+        
+        ci_label = round(( intervals * 100 ), ndigits = 2)
+        results.set_axis(['ES', str(ci_label) + '% CI'], axis = 1, inplace = True)     
+    
+    return results