diffindiff 2.2.6__tar.gz → 2.2.7__tar.gz

{diffindiff-2.2.6 → diffindiff-2.2.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: diffindiff
-Version: 2.2.6
+Version: 2.2.7
 Summary: diffindiff: Python library for convenient Difference-in-Differences analyses
 Author: Thomas Wieland
 Author-email: geowieland@googlemail.com
@@ -27,7 +27,7 @@ Thomas Wieland [ORCID](https://orcid.org/0000-0001-5168-9846) [EMail](mailto:geo
 
 If you use this software, please cite:
 
-Wieland, T. (2026). diffindiff: A Python library for convenient difference-in-differences analyses (Version 2.2.6) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.18656820
+Wieland, T. (2026). diffindiff: A Python library for convenient difference-in-differences analyses (Version 2.2.7) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.18656820
 
 
 ## Installation
@@ -167,11 +167,9 @@ See the /tests directory for usage examples of most of the included functions.
   - Wooldridge JM (2012) *Introductory Econometrics. A Modern Approach*.
 
 
-## What's new (v2.2.6)
+## What's new (v2.2.7)
+- Functions
+  - diddata.DiffData.define_treatment() for constructing a new treatment from a column in the dataframe 
 - Bugfixes:
-  - Check for correct dates in diddata.create_treatment()
-  - Check for valid columns in diddata.merge_data()
-  - Removed unnecessary old dependencies and imports
-- Other:
-  - Changed diddata.DiffGroups.add_segmentation() to return a message rather than raising an exception when the DiffGroups object already includes a benefit group
-
+  - didtools.treatment_times() and didtools.is_multiple_treatment_period() now also identify continuous treatments correctly
+  - Fixed problematic type conversion in didtools.fit_metrics()

{diffindiff-2.2.6 → diffindiff-2.2.7}/README.md

@@ -19,7 +19,7 @@ Thomas Wieland [ORCID](https://orcid.org/0000-0001-5168-9846) [EMail](mailto:geo
 
 If you use this software, please cite:
 
-Wieland, T. (2026). diffindiff: A Python library for convenient difference-in-differences analyses (Version 2.2.6) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.18656820
+Wieland, T. (2026). diffindiff: A Python library for convenient difference-in-differences analyses (Version 2.2.7) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.18656820
 
 
 ## Installation
@@ -159,11 +159,9 @@ See the /tests directory for usage examples of most of the included functions.
   - Wooldridge JM (2012) *Introductory Econometrics. A Modern Approach*.
 
 
-## What's new (v2.2.6)
+## What's new (v2.2.7)
+- Functions
+  - diddata.DiffData.define_treatment() for constructing a new treatment from a column in the dataframe 
 - Bugfixes:
-  - Check for correct dates in diddata.create_treatment()
-  - Check for valid columns in diddata.merge_data()
-  - Removed unnecessary old dependencies and imports
-- Other:
-  - Changed diddata.DiffGroups.add_segmentation() to return a message rather than raising an exception when the DiffGroups object already includes a benefit group
-
+  - didtools.treatment_times() and didtools.is_multiple_treatment_period() now also identify continuous treatments correctly
+  - Fixed problematic type conversion in didtools.fit_metrics()

{diffindiff-2.2.6 → diffindiff-2.2.7}/diffindiff/config.py

@@ -4,23 +4,25 @@
 # Author:      Thomas Wieland 
 #              ORCID: 0000-0001-5168-9846
 #              mail: geowieland@googlemail.com              
-# Version:     1.0.5
-# Last update: 2026-02-20 17:30
+# Version:     1.0.6
+# Last update: 2026-02-26 18:04
 # Copyright (c) 2025-2026 Thomas Wieland
 #-----------------------------------------------------------------------
 
 # Basic config:
 
 PACKAGE_NAME = "diffindiff"
-PACKAGE_VERSION = "2.2.6"
+PACKAGE_VERSION = "2.2.7"
 
-VERBOSE = False
+VERBOSE = True
 
 ROUND_STATISTIC = 3
 ROUND_PERCENT = 2
 
 AUTO_SWITCH_TO_PREPOST = True
 
+ACCEPT_CONTINUOUS_TREATMENTS = True
+
 # Description texts:
 
 DID_DESCRIPTION = "Difference-in-Differences Analysis"

{diffindiff-2.2.6 → diffindiff-2.2.7}/diffindiff/didanalysis.py

@@ -4,8 +4,8 @@
 # Author:      Thomas Wieland 
 #              ORCID: 0000-0001-5168-9846
 #              mail: geowieland@googlemail.com              
-# Version:     2.2.3
-# Last update: 2026-02-20 17:42
+# Version:     2.2.4
+# Last update: 2026-02-26 18:04
 # Copyright (c) 2024-2026 Thomas Wieland
 #-----------------------------------------------------------------------
 
@@ -1356,8 +1356,8 @@ def did_analysis(
     missing_replace_by_zero: bool = False,
     fit_by = "ols_fit",
     verbose: bool = config.VERBOSE
-    ):
-    
+    ):    
+   
     tools.check_columns(
         df = data,
         columns = [
@@ -1384,6 +1384,12 @@ def did_analysis(
         verbose = verbose
         )
     
+    tools.is_numeric(
+        df = data,
+        columns = treatment_col,
+        verbose = verbose
+        )
+    
     cols_relevant = [
         unit_col,
         time_col,
@@ -1807,7 +1813,7 @@ def did_analysis(
         }
     
     if bonferroni:
-        confint_alpha = confint_alpha/no_treatments   
+        confint_alpha = confint_alpha/no_treatments
 
     if fit_by == "ml":
         fit_result = helper.ml_fit(
@@ -1824,7 +1830,7 @@ def did_analysis(
             cluster_SE_by = cluster_SE_by,
             verbose = verbose
         )
-    
+           
     model_results = helper.extract_model_results(
         fit_result = fit_result,
         TG_col = TG_col,

{diffindiff-2.2.6 → diffindiff-2.2.7}/diffindiff/didanalysis_helper.py

@@ -4,8 +4,8 @@
 # Author:      Thomas Wieland 
 #              ORCID: 0000-0001-5168-9846
 #              mail: geowieland@googlemail.com              
-# Version:     1.0.6
-# Last update: 2025-02-20 17:38
+# Version:     1.0.7
+# Last update: 2025-02-26 18:02
 # Copyright (c) 2025-2026 Thomas Wieland
 #-----------------------------------------------------------------------
 
@@ -203,7 +203,7 @@ def create_spillover(
                 time_col = time_col,
                 treatment_col = treatment,
                 create_TT_col = TT_col,
-                verbose = verbose
+                verbose = False
                 )[0]          
 
         sp_unit_col = f"{config.SPILLOVER_UNIT_PREFIX}{config.DELIMITER}{treatment}"
@@ -396,7 +396,11 @@ def treatment_diagnostics(
         )
     
     if verbose:
-        print(f"There are {no_treatments} treatments (simultaneous: {no_treatments-staggered_count}, staggered: {staggered_count}) with {untreated[0]} treated and {untreated[1]} untreated units.")
+        
+        if no_treatments > 1:
+            print(f"There are {no_treatments} treatments (simultaneous: {no_treatments-staggered_count}, staggered: {staggered_count}) with {untreated[0]} treated and {untreated[1]} untreated units.")
+        else:
+            print(f"There is {no_treatments} treatment (staggered: {staggered_count}) with {untreated[0]} treated and {untreated[1]} untreated units.")
 
     return [
         treatment_diagnostics_results,

{diffindiff-2.2.6 → diffindiff-2.2.7}/diffindiff/diddata.py

@@ -4,8 +4,8 @@
 # Author:      Thomas Wieland 
 #              ORCID: 0000-0001-5168-9846
 #              mail: geowieland@googlemail.com              
-# Version:     2.1.6
-# Last update: 2026-02-20 18:28
+# Version:     2.1.8
+# Last update: 2026-02-26 18:30
 # Copyright (c) 2024-2026 Thomas Wieland
 #-----------------------------------------------------------------------
 
@@ -486,7 +486,7 @@ class DiffData:
         variables: list = None,
         unit_col: str = None,
         time_col: str = None,
-        verbose: bool = config.VERBOSE
+        verbose: bool = False
         ):
         
         if unit_col is None and time_col is None:
@@ -579,6 +579,7 @@ class DiffData:
         
         self.data[0] = did_modeldata
         self.data[5] = variables
+        self.data[7][len(self.data[7])] = helper.create_timestamp(function="add_covariates")
 
         if verbose:
             print("OK")
@@ -622,7 +623,6 @@ class DiffData:
         groups_data_old = did_groups_old.get_data()
 
         did_modeldata_old = self.get_did_modeldata_df()
-        unit_id_col, time_col = self.get_unit_time_cols()
         outcome_col_original = self.data[3]
         unit_time_col_original = self.get_unit_time_cols()
         covariates = self.get_covariates()
@@ -728,21 +728,157 @@ class DiffData:
             timestamp = helper.create_timestamp(function="add_treatment")
             )
 
-        did_data_new = DiffData(
-            did_modeldata = did_modeldata_new, 
-            diff_groups = groups_new, 
-            diff_treatment = treatment_new, 
-            outcome_col_original = outcome_col_original,
-            unit_time_col_original = unit_time_col_original,
-            covariates = covariates,
-            treatment_cols = treatment_cols_new,
-            timestamp = helper.create_timestamp(function="add_segmentation")
+        if verbose:
+            print("OK")
+
+        self.data[0] = did_modeldata_new
+        self.data[1] = groups_new
+        self.data[2] = treatment_new
+        self.data[3] = outcome_col_original
+        self.data[4] = unit_time_col_original
+        self.data[5] = covariates
+        self.data[6] = treatment_cols_new
+        self.data[7][len(self.data[7])] = helper.create_timestamp(function="add_treatment")
+        
+        return self
+
+    def define_treatment(
+        self,
+        treatment_name,
+        after_treatment_period: bool = False,
+        after_treatment_name = None,
+        verbose: bool = config.VERBOSE
+        ):
+
+        if not treatment_name:            
+            raise ValueError("When adding a treatment from the data, you need to specify a treatment column with parameter treament_name = [your_treatment].")
+        
+        if treatment_name not in self.get_did_modeldata_df().columns:
+            raise KeyError(f"Column '{treatment_name}' not in data frame")
+
+        did_treatment_old = self.get_did_treatment()
+        treatment_config_old = did_treatment_old.get_config()
+        treatment_meta_old = did_treatment_old.get_metadata()        
+        no_treatments_old = treatment_meta_old["no_treatments"]
+
+        did_groups_old = self.get_did_groups()
+        groups_config_old = did_groups_old.get_config()
+        groups_data_old = did_groups_old.get_data()
+
+        did_modeldata_old = self.get_did_modeldata_df()
+        outcome_col_original = self.data[3]
+        unit_time_col_original = self.get_unit_time_cols()
+        covariates = self.get_covariates()
+
+        treatment_cols = self.get_treatment_cols()
+        treatment_cols_new = treatment_cols
+
+        no_treatments = no_treatments_old+1
+        key_counter = no_treatments-1        
+        
+        tt = tools.treatment_times(
+            data = did_modeldata_old,
+            unit_col=config.UNIT_COL,
+            time_col=config.TIME_COL,
+            treatment_col=treatment_name,
+            verbose=verbose
+        )
+        
+        tt_date = [datetime.strptime(t, treatment_meta_old["date_format"]) for t in tt[1]]
+        treatment_period_start = min(tt_date)
+        treatment_period_end = max(tt_date)
+        treatment_period_start = treatment_period_start.strftime("%Y-%m-%d")
+        treatment_period_end = treatment_period_end.strftime("%Y-%m-%d")        
+        
+        is_notreatment_result = tools.is_notreatment(
+            data = did_modeldata_old,
+            unit_col=config.UNIT_COL,
+            treatment_col=treatment_name,
+            verbose = verbose 
+            )
+        
+        treatment_group = is_notreatment_result[1]
+        control_group = is_notreatment_result[2]
+        
+        if verbose:
+            print(f"Constructing treatment from column '{treatment_name}'", end = " ... ")
+
+        new_groups = create_groups(
+            treatment_group = treatment_group, 
+            control_group = control_group,
+            treatment_name = treatment_name,
+            verbose=False
             )
+        new_groups_data_df = new_groups.get_data()[0]        
+        new_groups_config = new_groups.get_config()
+        TG_col = new_groups_config[0]["TG_col"]
+
+        new_treatment = create_treatment(
+            study_period = [treatment_meta_old["study_period_start"], treatment_meta_old["study_period_end"]],
+            treatment_period = [treatment_period_start, treatment_period_end],
+            freq = treatment_meta_old["frequency"],
+            date_format = treatment_meta_old["date_format"],
+            treatment_name = treatment_name,
+            pre_post = treatment_meta_old["pre_post"],
+            after_treatment_period = after_treatment_period,
+            verbose=False
+            )
+        
+        new_treatment_data_df = new_treatment.get_data()
+        
+        new_treatment_config = new_treatment.get_config()        
+        TT_col = new_treatment_config[0]["TT_col"]
+        ATT_col = new_treatment_config[0]["ATT_col"]
+       
+        treatment_cols_new[key_counter] = {
+            "TT_col": TT_col, 
+            "ATT_col": ATT_col, 
+            "treatment_name": treatment_name, 
+            "after_treatment_name": after_treatment_name                
+            }
+        
+        groups_config_new = groups_config_old
+        groups_config_new[key_counter] = new_groups_config[0] 
+        groups_data_new = groups_data_old
+        groups_data_old.append(new_groups_data_df) 
+        groups_new = DiffGroups(
+            groups_data_new, 
+            groups_config_new,
+            timestamp = helper.create_timestamp(function="define_treatment")
+            )
+
+        treatment_meta_new = treatment_meta_old
+        treatment_meta_new["no_treatments"] = no_treatments
+        treatment_config_new = treatment_config_old
+        treatment_config_new[key_counter] = new_treatment_config[0]
+        
+        treatment_new = DiffTreatment(
+            new_treatment_data_df, 
+            treatment_config_new,
+            treatment_meta_new,
+            timestamp = helper.create_timestamp(function="define_treatment")
+            )       
 
         if verbose:
             print("OK")
-
-        return did_data_new
+            
+        if treatment_name in covariates:
+            
+            if verbose:
+                print(f"NOTE: Column '{treatment_name}' was defined as covariate before and is now removed from covariates list.")
+            
+            covariates.remove(treatment_name)
+
+        self.data[0] = did_modeldata_old
+        self.data[1] = groups_new
+        self.data[2] = treatment_new
+        self.data[3] = outcome_col_original
+        self.data[4] = unit_time_col_original
+        self.data[5] = covariates
+        self.data[6] = treatment_cols_new
+        self.data[7][len(self.data[7])] = helper.create_timestamp(function="define_treatment")        
+        
+        return self
 
     def add_segmentation(
         self,
@@ -979,8 +1115,8 @@ class DiffData:
                 if value["after_treatment_name"] is not None:
                     after_treatment_col[key] = value["after_treatment_name"]
                 if value["ATT_col"] is not None:
-                    ATT_col[key] = value["ATT_col"]
-                    
+                    ATT_col[key] = value["ATT_col"]            
+            
             did_results = didanalysis.did_analysis(
                 data = did_modeldata,
                 TG_col = TG_col,
@@ -1038,9 +1174,6 @@ def merge_data(
             ]
         )
 
-    if verbose:
-        print("Merging groups and treatment data", end = " ... ")
-
     groups_data_df = diff_groups.get_data()
     groups_data_df = groups_data_df[0]
     
@@ -1096,6 +1229,9 @@ def merge_data(
         verbose=verbose
         )
 
+    if verbose:
+        print("Merging groups and treatment data", end = " ... ")
+
     if keep_columns:
         outcome_data_short = outcome_data
     else:
@@ -1129,7 +1265,8 @@ def merge_data(
             }
         }
 
-    timestamp = helper.create_timestamp(function="merge_data")
+    timestamp = {}
+    timestamp[0] = helper.create_timestamp(function="merge_data")
 
     did_data_all = DiffData(
         did_modeldata, 
@@ -1196,8 +1333,6 @@ def create_data(
         verbose = verbose
         )
 
-    did_data_all.timestamp = helper.create_timestamp(function="create_data")
-
     return did_data_all
 
 def create_counterfactual(

{diffindiff-2.2.6 → diffindiff-2.2.7}/diffindiff/didtools.py

@@ -4,8 +4,8 @@
 # Author:      Thomas Wieland 
 #              ORCID: 0000-0001-5168-9846
 #              mail: geowieland@googlemail.com              
-# Version:     2.1.5
-# Last update: 2026-02-20 17:43
+# Version:     2.1.6
+# Last update: 2026-02-26 18:33
 # Copyright (c) 2025-2026 Thomas Wieland
 #-----------------------------------------------------------------------
 
@@ -46,6 +46,30 @@ def check_columns(
         if missing_columns:
             raise KeyError(f"Data do not contain column(s): {', '.join(missing_columns)}")
 
+def is_numeric(
+    df: pd.DataFrame, 
+    columns: list,
+    verbose: bool = config.VERBOSE
+    ):
+    
+    if len(columns) > 0:
+        
+        if verbose:
+            print(f"Checking if column(s) {', '.join(columns)} are numeric", end=" ... ")
+        
+        non_numeric_columns = []
+        
+        for col in columns:
+        
+            if not pd.api.types.is_numeric_dtype(df[col]):
+                non_numeric_columns.append(col)
+        
+        if verbose:
+            print("OK")
+        
+        if non_numeric_columns:
+            raise KeyError(f"Data contain non-numeric column(s): {', '.join(non_numeric_columns)}")
+
 def panel_index(
     data: pd.DataFrame,
     unit_col: str,
@@ -527,8 +551,11 @@ def is_multiple_treatment_period(
         unit_treatment = data_sub[treatment_col]
 
         groups = (unit_treatment != unit_treatment.shift()).cumsum()
-
-        periods_count = (unit_treatment == 1).groupby(groups).any().sum()
+        
+        if config.ACCEPT_CONTINUOUS_TREATMENTS:
+            periods_count = (unit_treatment > 0).groupby(groups).any().sum()
+        else:
+            periods_count = (unit_treatment == 1).groupby(groups).any().sum()
 
         unit_treatment_periods[unit] = int(periods_count)
 
@@ -636,25 +663,31 @@ def treatment_times(
         verbose=verbose
         )
     
-    is_multiple_treatment_period(
+    is_multiple_treatment_period_result = is_multiple_treatment_period(
         data = data,
         unit_col = unit_col,
         treatment_col = treatment_col,
         verbose = verbose
-        )[0]
+        )    
     
     if verbose:
         print(f"Identifying treatment times for treatment '{treatment_col}'", end = " ... ")
     
-    tt = list(unique(data.loc[data[treatment_col] == 1, time_col]))
-
+    if config.ACCEPT_CONTINUOUS_TREATMENTS:
+        tt = list(unique(data.loc[data[treatment_col] > 0, time_col]))
+    else:
+        tt = list(unique(data.loc[data[treatment_col] == 1, time_col]))
+    
     units = unique(data[unit_col])
     
     units_tt = pd.DataFrame(columns = [unit_col, "treatment_min", "treatment_max"])
     
     for unit in units:
         
-        data_unit_tt = data[(data[unit_col] == unit) & (data[treatment_col] == 1)]
+        if config.ACCEPT_CONTINUOUS_TREATMENTS:
+            data_unit_tt = data[(data[unit_col] == unit) & (data[treatment_col] > 0)]
+        else:
+            data_unit_tt = data[(data[unit_col] == unit) & (data[treatment_col] == 1)]
         
         if data_unit_tt.empty:
             continue
@@ -678,7 +711,7 @@ def treatment_times(
         
     if verbose:
         print("OK")
-        
+            
     return [
         units_tt,
         tt
@@ -796,9 +829,9 @@ def fit_metrics(
 
     assert observed_no == expected_no, "Error while calculating fit metrics: Observed and expected differ in length"
     
-    if not pd.api.types.is_numeric_dtype(observed):
+    if not pd.api.types.is_numeric_dtype(observed) or not np.issubdtype(observed.dtype, np.number):
         raise ValueError("Error while calculating fit metrics: Observed column is not numeric")
-    if not pd.api.types.is_numeric_dtype(expected):
+    if not pd.api.types.is_numeric_dtype(expected) or not np.issubdtype(expected.dtype, np.number):
         raise ValueError("Error while calculating fit metrics: Expected column is not numeric")
     
     if outcome_col is not None:
@@ -810,8 +843,8 @@ def fit_metrics(
     
     if remove_nan:
         
-        observed = observed.reset_index(drop=True)
-        expected = expected.reset_index(drop=True)
+        observed = np.array(observed)
+        expected = np.array(expected)
 
         obs_exp = pd.DataFrame(
             {
@@ -968,6 +1001,7 @@ def check_date_format(
     
     if len(invalid_dates) > 0:
         invalid_dates_included = True
+        invalid_dates = [str(d) for d in invalid_dates]
     
     return [
         invalid_dates_included,

{diffindiff-2.2.6 → diffindiff-2.2.7}/diffindiff.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: diffindiff
-Version: 2.2.6
+Version: 2.2.7
 Summary: diffindiff: Python library for convenient Difference-in-Differences analyses
 Author: Thomas Wieland
 Author-email: geowieland@googlemail.com
@@ -27,7 +27,7 @@ Thomas Wieland [ORCID](https://orcid.org/0000-0001-5168-9846) [EMail](mailto:geo
 
 If you use this software, please cite:
 
-Wieland, T. (2026). diffindiff: A Python library for convenient difference-in-differences analyses (Version 2.2.6) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.18656820
+Wieland, T. (2026). diffindiff: A Python library for convenient difference-in-differences analyses (Version 2.2.7) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.18656820
 
 
 ## Installation
@@ -167,11 +167,9 @@ See the /tests directory for usage examples of most of the included functions.
   - Wooldridge JM (2012) *Introductory Econometrics. A Modern Approach*.
 
 
-## What's new (v2.2.6)
+## What's new (v2.2.7)
+- Functions
+  - diddata.DiffData.define_treatment() for constructing a new treatment from a column in the dataframe 
 - Bugfixes:
-  - Check for correct dates in diddata.create_treatment()
-  - Check for valid columns in diddata.merge_data()
-  - Removed unnecessary old dependencies and imports
-- Other:
-  - Changed diddata.DiffGroups.add_segmentation() to return a message rather than raising an exception when the DiffGroups object already includes a benefit group
-
+  - didtools.treatment_times() and didtools.is_multiple_treatment_period() now also identify continuous treatments correctly
+  - Fixed problematic type conversion in didtools.fit_metrics()

{diffindiff-2.2.6 → diffindiff-2.2.7}/setup.py

@@ -7,7 +7,7 @@ def read_README():
     
 setup(
     name='diffindiff',
-    version='2.2.6',
+    version='2.2.7',
     description='diffindiff: Python library for convenient Difference-in-Differences analyses',
     packages=find_packages(include=["diffindiff", "diffindiff.tests"]),
     include_package_data=True,