Merge pull request #64 from MITLibraries/HRQB-34-base-salary-change-p…

…ercent

HRQB 34 - Add base salary change percent

hrqb/tasks/employee_salary_history.py

@@ -30,23 +30,9 @@ def requires(self) -> list[luigi.Task]:  # pragma: nocover
 
     def get_dataframe(self) -> pd.DataFrame:
         dw_salary_df = self.named_inputs["ExtractDWEmployeeSalaryHistory"].read()
-        qb_emp_appts_df = self.named_inputs["ExtractQBEmployeeAppointments"].read()
 
         # merge with employee appointment data for QB appointment record identifier
-        qb_emp_appts_df = qb_emp_appts_df[
-            [
-                "Record ID#",
-                "HR Appointment Key",
-                "Begin Date",
-                "End Date",
-            ]
-        ].rename(
-            columns={
-                "Record ID#": "related_employee_appointment_id",
-                "Begin Date": "appointment_begin_date",
-                "End Date": "appointment_end_date",
-            }
-        )
+        qb_emp_appts_df = self._get_employee_appointments()
         salary_df = dw_salary_df.merge(
             qb_emp_appts_df,
             how="left",
@@ -63,6 +49,9 @@ def get_dataframe(self) -> pd.DataFrame:
         salary_df["original_effort"] = salary_df["original_effort"] / 100.0
         salary_df["temp_effort"] = salary_df["temp_effort"] / 100.0
 
+        # set base salary change percentage from previous record, for same position
+        salary_df = self._set_base_salary_change_percent(salary_df)
+
         # mint a unique, deterministic value for the merge "Key" field
         salary_df["key"] = salary_df.apply(
             lambda row: md5_hash_from_values(
@@ -90,13 +79,65 @@ def get_dataframe(self) -> pd.DataFrame:
             "original_base_amount": "Base Salary",
             "original_hourly_rate": "Hourly",
             "original_effort": "Effort %",
+            "base_change_percent": "Salary Change %",
             "temp_change_base_amount": "Temp Base Salary",
             "temp_change_hourly_rate": "Temp Hourly",
             "temp_effort": "Temp Effort %",
             "key": "Key",
         }
         return salary_df[fields.keys()].rename(columns=fields)
 
+    def _get_employee_appointments(self) -> pd.DataFrame:
+        qb_emp_appts_df = self.named_inputs["ExtractQBEmployeeAppointments"].read()
+        return qb_emp_appts_df[
+            [
+                "Record ID#",
+                "HR Appointment Key",
+                "Begin Date",
+                "End Date",
+            ]
+        ].rename(
+            columns={
+                "Record ID#": "related_employee_appointment_id",
+                "Begin Date": "appointment_begin_date",
+                "End Date": "appointment_end_date",
+            }
+        )
+
+    def _set_base_salary_change_percent(self, salary_df: pd.DataFrame) -> pd.DataFrame:
+        """Create column with percentage change between sequential salaries.
+
+        This method:
+            1. sorts by appointment MIT ID and appointment dates
+            2. groups the salary dataframe by MIT ID and unique appointment identifier
+            3. select the base salary from the PREVIOUS salary
+            4. calculates percentage change
+        """
+        new_salary_df = salary_df.copy()
+        new_salary_df["previous_base_amount"] = (
+            new_salary_df.sort_values(
+                [
+                    "mit_id",
+                    "appointment_begin_date",
+                    "appointment_end_date",
+                ]
+            )
+            .groupby(["mit_id", "hr_appt_key"])["original_base_amount"]
+            .shift(1)
+        )
+        new_salary_df["base_change_percent"] = round(
+            (
+                new_salary_df["original_base_amount"]
+                / new_salary_df["previous_base_amount"]
+                - 1.0
+            ),
+            3,
+        )
+        new_salary_df["base_change_percent"] = new_salary_df["base_change_percent"].where(
+            new_salary_df["previous_base_amount"].notna(), 0.0
+        )
+        return new_salary_df
+
 
 class LoadEmployeeSalaryHistory(QuickbaseUpsertTask):
     table_name = luigi.Parameter("Employee Salary History")

tests/tasks/test_employee_salary_history.py

@@ -1,4 +1,7 @@
-# ruff: noqa: PLR2004, PD901
+# ruff: noqa: PLR2004, PD901, SLF001
+
+import numpy as np
+import pandas as pd
 
 from hrqb.utils import md5_hash_from_values
 
@@ -60,3 +63,52 @@ def test_task_transform_employee_salary_history_key_expected_from_input_data(
             emp_salary_row["End Date"],
         ]
     )
+
+
+def test_task_transform_employee_salary_history_set_base_change_percent(
+    task_transform_employee_salary_history_complete,
+):
+    df = pd.DataFrame(
+        [
+            ("123456789", "123", "2020-01-01", "2021-06-30", 10_000),
+            ("123456789", "123", "2020-07-01", "2021-12-31", 10_300),
+            ("123456789", "123", "2021-01-01", "2021-06-30", 15_000),
+            ("123456789", "456", "2021-07-01", "2022-06-30", 20_000),
+            ("123456789", "456", "2022-07-01", "2022-12-31", 22_500),
+            ("123456789", "456", "2023-01-01", "2999-12-31", 24_750),
+        ],
+        columns=[
+            "mit_id",
+            "hr_appt_key",
+            "appointment_begin_date",
+            "appointment_end_date",
+            "original_base_amount",
+        ],
+    )
+    new_df = (
+        task_transform_employee_salary_history_complete._set_base_salary_change_percent(
+            df
+        )
+    )
+    np.testing.assert_array_equal(
+        new_df.previous_base_amount.values,
+        [
+            np.nan,  # first position, so no previous salary
+            10_000.0,
+            10_300.0,
+            np.nan,  # new position, so previous salary None
+            20_000.0,
+            22_500.0,
+        ],
+    )
+    np.testing.assert_array_equal(
+        new_df.base_change_percent.values,
+        [
+            0.0,  # first position, so no change
+            0.03,
+            0.456,
+            0.0,  # new position, so no change
+            0.125,
+            0.1,
+        ],
+    )