feat: add unit tests for core functions

tests/test_unit.py

@@ -0,0 +1,157 @@
+import pandas as pd
+import pytest
+from app.main import depart, inconsistency, promotion, developpement, interpret_shap
+import os
+
+
+# ---------------------------------------------------------------------------
+# depart
+# ---------------------------------------------------------------------------
+
+def test_depart():
+    assert depart(0) == "The staff has a LOW probability of resigning"
+    assert depart(1) == "The staff has a HIGH probability of resigning"
+
+# ---------------------------------------------------------------------------
+# inconsistency
+# The function is called via df.apply(inconsistency, axis=1), so each input
+# is a pandas Series representing a single row.
+# ---------------------------------------------------------------------------
+
+class TestInconsistency:
+    # --- Commercial ---
+    def test_commercial_marketing_is_consistent(self):
+        row = pd.Series({"departement": "Commercial", "domaine_etude": "Marketing"})
+        assert inconsistency(row) == 0
+
+    def test_commercial_infra_is_inconsistent(self):
+        row = pd.Series({"departement": "Commercial", "domaine_etude": "Infra & Cloud"})
+        assert inconsistency(row) == 1
+
+    def test_commercial_rh_is_inconsistent(self):
+        row = pd.Series({"departement": "Commercial", "domaine_etude": "Ressources Humaines"})
+        assert inconsistency(row) == 1
+
+    def test_commercial_transformation_is_inconsistent(self):
+        row = pd.Series({"departement": "Commercial", "domaine_etude": "Transformation Digitale"})
+        assert inconsistency(row) == 1
+
+    # --- Consulting ---
+    def test_consulting_infra_is_consistent(self):
+        row = pd.Series({"departement": "Consulting", "domaine_etude": "Infra & Cloud"})
+        assert inconsistency(row) == 0
+
+    def test_consulting_transformation_is_consistent(self):
+        row = pd.Series({"departement": "Consulting", "domaine_etude": "Transformation Digitale"})
+        assert inconsistency(row) == 0
+
+    def test_consulting_marketing_is_inconsistent(self):
+        row = pd.Series({"departement": "Consulting", "domaine_etude": "Marketing"})
+        assert inconsistency(row) == 1
+
+    def test_consulting_rh_is_inconsistent(self):
+        row = pd.Series({"departement": "Consulting", "domaine_etude": "Ressources Humaines"})
+        assert inconsistency(row) == 1
+
+    # --- Ressources Humaines ---
+    def test_rh_rh_is_consistent(self):
+        row = pd.Series({"departement": "Ressources Humaines", "domaine_etude": "Ressources Humaines"})
+        assert inconsistency(row) == 0
+
+    def test_rh_entrepreneuriat_is_consistent(self):
+        row = pd.Series({"departement": "Ressources Humaines", "domaine_etude": "Entrepreunariat"})
+        assert inconsistency(row) == 0
+
+    def test_rh_marketing_is_inconsistent(self):
+        row = pd.Series({"departement": "Ressources Humaines", "domaine_etude": "Marketing"})
+        assert inconsistency(row) == 1
+
+    def test_rh_infra_is_inconsistent(self):
+        row = pd.Series({"departement": "Ressources Humaines", "domaine_etude": "Infra & Cloud"})
+        assert inconsistency(row) == 1
+
+
+# ---------------------------------------------------------------------------
+# promotion
+# Returns 1 (stagnant) only when BOTH values are strictly greater than 4.
+# ---------------------------------------------------------------------------
+
+class TestPromotion:
+    def test_both_above_4_is_stagnant(self):
+        row = pd.Series({"annes_sous_responsable_actuel": 5, "annees_depuis_la_derniere_promotion": 5})
+        assert promotion(row) == 1
+
+    def test_large_values_are_stagnant(self):
+        row = pd.Series({"annes_sous_responsable_actuel": 10, "annees_depuis_la_derniere_promotion": 10})
+        assert promotion(row) == 1
+
+    def test_manager_years_exactly_4_not_stagnant(self):
+        row = pd.Series({"annes_sous_responsable_actuel": 4, "annees_depuis_la_derniere_promotion": 5})
+        assert promotion(row) == 0
+
+    def test_promotion_years_exactly_4_not_stagnant(self):
+        row = pd.Series({"annes_sous_responsable_actuel": 5, "annees_depuis_la_derniere_promotion": 4})
+        assert promotion(row) == 0
+
+    def test_both_below_threshold_not_stagnant(self):
+        row = pd.Series({"annes_sous_responsable_actuel": 2, "annees_depuis_la_derniere_promotion": 2})
+        assert promotion(row) == 0
+
+    def test_only_manager_years_high_not_stagnant(self):
+        row = pd.Series({"annes_sous_responsable_actuel": 6, "annees_depuis_la_derniere_promotion": 1})
+        assert promotion(row) == 0
+
+    def test_only_promotion_years_high_not_stagnant(self):
+        row = pd.Series({"annes_sous_responsable_actuel": 1, "annees_depuis_la_derniere_promotion": 6})
+        assert promotion(row) == 0
+
+
+# ---------------------------------------------------------------------------
+# developpement
+# Returns 0 for new employees (tenure == 0), 1 when tenure >= 2 with <= 1
+# training, 0 otherwise (sufficient training, or tenure == 1 year).
+# ---------------------------------------------------------------------------
+
+class TestDeveloppement:
+    def test_new_employee_returns_0(self):
+        row = pd.Series({"annees_dans_l_entreprise": 0, "nb_formations_suivies": 0})
+        assert developpement(row) == 0
+
+    def test_one_year_tenure_returns_0(self):
+        # tenure == 1 meets neither condition → 0
+        row = pd.Series({"annees_dans_l_entreprise": 1, "nb_formations_suivies": 0})
+        assert developpement(row) == 0
+
+    def test_two_years_no_training_is_stagnant(self):
+        row = pd.Series({"annees_dans_l_entreprise": 2, "nb_formations_suivies": 0})
+        assert developpement(row) == 1
+
+    def test_two_years_one_training_is_stagnant(self):
+        row = pd.Series({"annees_dans_l_entreprise": 2, "nb_formations_suivies": 1})
+        assert developpement(row) == 1
+
+    def test_two_years_two_trainings_not_stagnant(self):
+        row = pd.Series({"annees_dans_l_entreprise": 2, "nb_formations_suivies": 2})
+        assert developpement(row) == 0
+
+    def test_long_tenure_no_training_is_stagnant(self):
+        row = pd.Series({"annees_dans_l_entreprise": 10, "nb_formations_suivies": 0})
+        assert developpement(row) == 1
+
+    def test_long_tenure_sufficient_training_not_stagnant(self):
+        row = pd.Series({"annees_dans_l_entreprise": 10, "nb_formations_suivies": 5})
+        assert developpement(row) == 0
+
+
+def test_interpret_shap():
+    assert interpret_shap(0, 0.5) == "Primary driver — increases resignation risk"
+    assert interpret_shap(0, -0.5) == "Primary driver — decreases resignation risk"
+    assert interpret_shap(1, 0.5) == "Strong factor — increases resignation risk"
+    assert interpret_shap(1, -0.5) == "Strong factor — decreases resignation risk"
+    assert interpret_shap(2, 0.5) == "Moderate factor — increases resignation risk"
+    assert interpret_shap(2, -0.5) == "Moderate factor — decreases resignation risk"
+    assert interpret_shap(3, 0.5) == "Contributing factor — increases resignation risk"
+    assert interpret_shap(3, -0.5) == "Contributing factor — decreases resignation risk"
+    assert interpret_shap(4, 0.5) == "Notable factor — increases resignation risk"
+    assert interpret_shap(4, -0.5) == "Notable factor — decreases resignation risk"
+