Spp 5901 rename methods

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "statistical_methods_library"
-version = "4.5.0"
+version = "5.0.0"
 description = ""
 authors = ["Your Name <[email protected]>"]
 license = "MIT"

statistical_methods_library/estimation.py

@@ -16,7 +16,7 @@ class ValidationError(Exception):
     pass
 
 
-def estimate(
+def ht_ratio(
     input_df: DataFrame,
     unique_identifier_col: str,
     period_col: str,
@@ -33,8 +33,8 @@ def estimate(
     calibration_weight_col: typing.Optional[str] = "calibration_weight",
 ) -> DataFrame:
     """
-    Perform estimation of design and calibration weights using Expansion and
-    Ratio estimation.
+    Perform Horvitz-Thompson estimation of design and calibration weights
+    using Expansion and Ratio estimation.
 
     ###Arguments
     * `input_df`: The input data frame.

statistical_methods_library/imputation.py

@@ -56,7 +56,7 @@ class DataIntegrityError(ImputationError):
     pass
 
 
-def impute(
+def ratio_of_means(
     input_df: DataFrame,
     reference_col: str,
     period_col: str,

statistical_methods_library/winsorisation.py → statistical_methods_library/outliering.py

@@ -31,7 +31,7 @@ class Marker(Enum):
     """The value has not been winsorised because design * calibration is <= 1."""
 
 
-def one_sided_winsorise(
+def winsorise(
     input_df: DataFrame,
     reference_col: str,
     period_col: str,
@@ -43,7 +43,7 @@ def one_sided_winsorise(
     calibration_col: typing.Optional[str] = None,
     auxiliary_col: typing.Optional[str] = None,
     marker_col: typing.Optional[str] = "winsorisation_marker",
-):
+) -> DataFrame:
     """
     Perform One-sided Winsorisation.
 

tests/conftest.py

@@ -11,13 +11,14 @@ def fxt_load_test_csv(fxt_spark_session):
     # be called as a function in our tests and thus this fixture would have no
     # access to its return value. We do this so that our loader can be
     # passed a file name.
-    def load(columns, types, test_module, test_category, test_data):
+    def load(columns, types, test_module, test_function, test_category, test_data):
         test_dataframe = fxt_spark_session.read.csv(
             str(
                 pathlib.Path(
                     "tests",
                     "fixture_data",
                     test_module,
+                    test_function,
                     test_category,
                     f"{test_data}.csv",
                 )

tests/fixture_data/imputation/back_data_scenarios/back_data_input.csv → tests/fixture_data/imputation/ratio_of_means/back_data_scenarios/back_data_input.csv

@@ -1,23 +1,23 @@
-reference,period,strata,target,auxiliary,
-1111000,203902,9000,6000,2,
-1111001,203902,8000,7000,2,
-1111002,203902,9000,,2,
-1111003,203902,8000,,2,
-1111004,203902,9000,,2,
-1111007,203902,8000,,2,
-1111000,203903,9000,,2,
-1111001,203903,8000,,2,
-1111002,203903,9000,1500.2,2,
-1111003,203903,8000,2000,2,
-1111004,203903,9000,,2,
-1111005,203903,8000,1220,2,
-1111006,203903,9000,4500,2,
-1111007,203903,8000,,2,
-1111000,203904,9000,,2,
-1111001,203904,8000,,2,
-1111002,203904,9000,,2,
-1111003,203904,8000,1000,2,
-1111004,203904,9000,2300,2,
-1111005,203904,8000,,2,
-1111006,203904,9000,7000,2,
+reference,period,strata,target,auxiliary,
+1111000,203902,9000,6000,2,
+1111001,203902,8000,7000,2,
+1111002,203902,9000,,2,
+1111003,203902,8000,,2,
+1111004,203902,9000,,2,
+1111007,203902,8000,,2,
+1111000,203903,9000,,2,
+1111001,203903,8000,,2,
+1111002,203903,9000,1500.2,2,
+1111003,203903,8000,2000,2,
+1111004,203903,9000,,2,
+1111005,203903,8000,1220,2,
+1111006,203903,9000,4500,2,
+1111007,203903,8000,,2,
+1111000,203904,9000,,2,
+1111001,203904,8000,,2,
+1111002,203904,9000,,2,
+1111003,203904,8000,1000,2,
+1111004,203904,9000,2300,2,
+1111005,203904,8000,,2,
+1111006,203904,9000,7000,2,
 1111007,203904,8000,,2,

tests/fixture_data/imputation/back_data_scenarios/back_data_missing_strata_input.csv → tests/fixture_data/imputation/ratio_of_means/back_data_scenarios/back_data_missing_strata_input.csv

@@ -1,11 +1,11 @@
-reference,period,strata,target,auxiliary
-101,204001,9000,,2
-101,204002,9000,10000,2
-102,204002,7000,,2
-101,204003,9000,,2
-102,204003,7000,20000,2
-103,204003,9000,,2
-101,204004,9000,,2
-102,204004,7000,,2
-103,204004,9000,,2
+reference,period,strata,target,auxiliary
+101,204001,9000,,2
+101,204002,9000,10000,2
+102,204002,7000,,2
+101,204003,9000,,2
+102,204003,7000,20000,2
+103,204003,9000,,2
+101,204004,9000,,2
+102,204004,7000,,2
+103,204004,9000,,2
 104,204004,7000,9000,2

tests/fixture_data/imputation/dev_scenarios/divide_by_zero_input.csv → tests/fixture_data/imputation/ratio_of_means/dev_scenarios/divide_by_zero_input.csv

@@ -1,17 +1,17 @@
-reference,period,strata,target,auxiliary
-1,202001,1,10,1
-1,202002,1,20,1
-1,202003,1,,1
-1,202004,1,,1
-2,202001,1,10,1
-2,202002,1,20,1
-2,202003,1,0,1
-2,202004,1,30,1
-3,202001,1,10,1
-3,202002,1,,1
-3,202003,1,20,1
-3,202004,1,,1
-4,202001,1,10,1
-4,202002,1,,1
-4,202003,1,,1
-4,202004,1,,1
+reference,period,strata,target,auxiliary
+1,202001,1,10,1
+1,202002,1,20,1
+1,202003,1,,1
+1,202004,1,,1
+2,202001,1,10,1
+2,202002,1,20,1
+2,202003,1,0,1
+2,202004,1,30,1
+3,202001,1,10,1
+3,202002,1,,1
+3,202003,1,20,1
+3,202004,1,,1
+4,202001,1,10,1
+4,202002,1,,1
+4,202003,1,,1
+4,202004,1,,1

tests/fixture_data/imputation/methodology_scenarios/13_R_FI_FI_NS_BI_BI_R_input.csv → tests/fixture_data/imputation/ratio_of_means/methodology_scenarios/13_R_FI_FI_NS_BI_BI_R_input.csv

@@ -1,28 +1,28 @@
-reference,period,strata,target,auxiliary
-130001,202001,100,6583,20
-130001,202002,100,3759,20
-130001,202003,100,8256,20
-130001,202004,100,6743,20
-130001,202005,100,6811,20
-130001,202006,100,9896,20
-130001,202007,100,3282,20
-130002,202001,100,687,69
-130002,202002,100,8987,69
-130002,202003,100,371,69
-130002,202004,100,3385,69
-130002,202005,100,6606,69
-130002,202006,100,5901,69
-130002,202007,100,9834,69
-130003,202001,100,7113,39
-130003,202002,100,8583,39
-130003,202003,100,5129,39
-130003,202004,100,2024,39
-130003,202005,100,9422,39
-130003,202006,100,8710,39
-130003,202007,100,6461,39
-130004,202001,100,2571,37
-130004,202002,100,,37
-130004,202003,100,,37
-130004,202005,100,,37
-130004,202006,100,,37
-130004,202007,100,6621,37
+reference,period,strata,target,auxiliary
+130001,202001,100,6583,20
+130001,202002,100,3759,20
+130001,202003,100,8256,20
+130001,202004,100,6743,20
+130001,202005,100,6811,20
+130001,202006,100,9896,20
+130001,202007,100,3282,20
+130002,202001,100,687,69
+130002,202002,100,8987,69
+130002,202003,100,371,69
+130002,202004,100,3385,69
+130002,202005,100,6606,69
+130002,202006,100,5901,69
+130002,202007,100,9834,69
+130003,202001,100,7113,39
+130003,202002,100,8583,39
+130003,202003,100,5129,39
+130003,202004,100,2024,39
+130003,202005,100,9422,39
+130003,202006,100,8710,39
+130003,202007,100,6461,39
+130004,202001,100,2571,37
+130004,202002,100,,37
+130004,202003,100,,37
+130004,202005,100,,37
+130004,202006,100,,37
+130004,202007,100,6621,37

tests/fixture_data/imputation/methodology_scenarios/14_C_FI_FI_NS_BI_BI_R_input.csv → tests/fixture_data/imputation/ratio_of_means/methodology_scenarios/14_C_FI_FI_NS_BI_BI_R_input.csv

@@ -1,28 +1,28 @@
-reference,period,strata,target,auxiliary
-140001,202001,100,3582,27
-140001,202002,100,4533,27
-140001,202003,100,4471,27
-140001,202004,100,3040,27
-140001,202005,100,8519,27
-140001,202006,100,8690,27
-140001,202007,100,9979,27
-140002,202001,100,5359,83
-140002,202002,100,7091,83
-140002,202003,100,7098,83
-140002,202004,100,7408,83
-140002,202005,100,1937,83
-140002,202006,100,6711,83
-140002,202007,100,1052,83
-140003,202001,100,439,89
-140003,202002,100,7783,89
-140003,202003,100,897,89
-140003,202004,100,8477,89
-140003,202005,100,2207,89
-140003,202006,100,9889,89
-140003,202007,100,2528,89
-140004,202001,100,,28
-140004,202002,100,,28
-140004,202003,100,,28
-140004,202005,100,,28
-140004,202006,100,,28
-140004,202007,100,7243,28
+reference,period,strata,target,auxiliary
+140001,202001,100,3582,27
+140001,202002,100,4533,27
+140001,202003,100,4471,27
+140001,202004,100,3040,27
+140001,202005,100,8519,27
+140001,202006,100,8690,27
+140001,202007,100,9979,27
+140002,202001,100,5359,83
+140002,202002,100,7091,83
+140002,202003,100,7098,83
+140002,202004,100,7408,83
+140002,202005,100,1937,83
+140002,202006,100,6711,83
+140002,202007,100,1052,83
+140003,202001,100,439,89
+140003,202002,100,7783,89
+140003,202003,100,897,89
+140003,202004,100,8477,89
+140003,202005,100,2207,89
+140003,202006,100,9889,89
+140003,202007,100,2528,89
+140004,202001,100,,28
+140004,202002,100,,28
+140004,202003,100,,28
+140004,202005,100,,28
+140004,202006,100,,28
+140004,202007,100,7243,28

tests/fixture_data/imputation/methodology_scenarios/15_BI_BI_R_NS_R_FI_FI_input.csv → tests/fixture_data/imputation/ratio_of_means/methodology_scenarios/15_BI_BI_R_NS_R_FI_FI_input.csv

@@ -1,28 +1,28 @@
-reference,period,strata,target,auxiliary
-150001,202001,100,5891,15
-150001,202002,100,4581,15
-150001,202003,100,4741,15
-150001,202004,100,8092,15
-150001,202005,100,2727,15
-150001,202006,100,2363,15
-150001,202007,100,9549,15
-150002,202001,100,8611,5
-150002,202002,100,9383,5
-150002,202003,100,1064,5
-150002,202004,100,7419,5
-150002,202005,100,9967,5
-150002,202006,100,1393,5
-150002,202007,100,2593,5
-150003,202001,100,3426,79
-150003,202002,100,6246,79
-150003,202003,100,8402,79
-150003,202004,100,3891,79
-150003,202005,100,5463,79
-150003,202006,100,6852,79
-150003,202007,100,2959,79
-150004,202001,100,,69
-150004,202002,100,,69
-150004,202003,100,4208,69
-150004,202005,100,2818,69
-150004,202006,100,,69
-150004,202007,100,,69
+reference,period,strata,target,auxiliary
+150001,202001,100,5891,15
+150001,202002,100,4581,15
+150001,202003,100,4741,15
+150001,202004,100,8092,15
+150001,202005,100,2727,15
+150001,202006,100,2363,15
+150001,202007,100,9549,15
+150002,202001,100,8611,5
+150002,202002,100,9383,5
+150002,202003,100,1064,5
+150002,202004,100,7419,5
+150002,202005,100,9967,5
+150002,202006,100,1393,5
+150002,202007,100,2593,5
+150003,202001,100,3426,79
+150003,202002,100,6246,79
+150003,202003,100,8402,79
+150003,202004,100,3891,79
+150003,202005,100,5463,79
+150003,202006,100,6852,79
+150003,202007,100,2959,79
+150004,202001,100,,69
+150004,202002,100,,69
+150004,202003,100,4208,69
+150004,202005,100,2818,69
+150004,202006,100,,69
+150004,202007,100,,69