Merge pull request #7 from lgabs/upgrade-pandas

Update pandas version

.github/workflows/ci.yml

@@ -0,0 +1,32 @@
+name: Python application test
+
+on:
+  push:
+    branches: [ master ]
+  pull_request:
+    branches: [ master ]
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        python-version: ['3.10']
+
+    steps:
+    - name: Checkout repository
+      uses: actions/checkout@v2
+
+    - name: Set up Python ${{ matrix.python-version }}
+      uses: actions/setup-python@v2
+      with:
+        python-version: ${{ matrix.python-version }}
+
+    - name: Install dependencies
+      run: |
+        python -m pip install --upgrade pip
+        pip install -r requirements.txt
+
+    - name: Run tests
+      run: |
+        python -m unittest discover tests

Makefile

@@ -6,7 +6,7 @@ install-lib:
 	python setup.py bdist_wheel
 	pip3 install --force-reinstall  dist/**.whl
 
-tests:
+test:
 	python -m unittest discover
 
 lint:

janus/stats/experiment.py

@@ -105,7 +105,6 @@ def calculate_probabilities(self, other_variant, metric_name: str):
     def calculate_bootstrap(
         self, metric_name: str, eval_function=get_bootstraped_mean, spark_session=None
     ):
-
         if metric_name == "revenue":
             data = list(self.variant_df.query("sales == 1")["revenue"])
         elif metric_name == "conversion":
@@ -143,7 +142,6 @@ def __init__(
         self.results = {}
 
     def run_experiment(self, df_results_per_user: pd.DataFrame):
-
         logging.info(f"INITIALIZING experiment '{self.name}' evaluation...")
         variant_names = list(df_results_per_user.alternative.unique())
         assert (
@@ -201,7 +199,6 @@ def evaluate_statistics(
             variant1.calculate_probabilities(variant2, metric_name)
 
     def consolidate_results(self, variantA: Variant, variantB: Variant):
-
         for variant_interest, variant_other in zip(
             (variantA, variantB), (variantB, variantA)
         ):

janus/utils/make_test_dataframe.py

@@ -11,7 +11,6 @@ def createTestDataFrame(
     average_ticket_baseline: float,
     average_ticket_alternative: float,
 ):
-
     # testes antes de começar função:
     # TODO: 1. nenhum parâmetro nulo
     # TODO: 2. todos os tipos estão certos

pages/1_Analyze_with_Summary_Information.py

@@ -40,10 +40,7 @@
         value=100,
     )
     control_total_value = st.number_input(
-        label="Total Conversion Value in Control",
-        value=200.0,
-        step=1.,
-        format="%.4f"
+        label="Total Conversion Value in Control", value=200.0, step=1.0, format="%.4f"
     )
 
     # Treatment
@@ -58,11 +55,13 @@
     test_total_value = st.number_input(
         label="Total Conversion Value in Treatment",
         value=250.00,
-        step=1.,
-        format="%.4f"
+        step=1.0,
+        format="%.4f",
     )
 
-    experiment_name = st.text_input(label='Experiment Name (Optional)', value='My Experiment')
+    experiment_name = st.text_input(
+        label="Experiment Name (Optional)", value="My Experiment"
+    )
 
     submit_button = st.form_submit_button(label="Run Experiment")
 
@@ -83,7 +82,7 @@
     df_per_user_simulated = create_per_user_dataframe_multivariate(
         df, conversion_value_cols=conversion_value_cols
     )
-    
+
     # st.write("df")
     # st.dataframe(df)
     # st.write("df_per_user_simulated")
@@ -95,16 +94,16 @@
         # TODO: generalize this code for all pages and generalize
         # lib's revenue col to monetary values
         df_per_user_simulated = df_per_user_simulated.rename(
-            columns={"converted": "sales", "total_value": 'revenue'}
+            columns={"converted": "sales", "total_value": "revenue"}
         )  # hacking, sales are generic conversions in janus lib
         experiment = Experiment(
             name=experiment_name,
             keymetrics=["conversion", "revenue", "arpu"],
-            baseline_variant_name='control',
+            baseline_variant_name="control",
         )
         experiment.run_experiment(df_results_per_user=df_per_user_simulated)
         save_results_in_session_state(
-            experiment, control_label='control', treatment_label='treatment'
+            experiment, control_label="control", treatment_label="treatment"
         )
 
         # Show Results in dataframe form v0
@@ -120,4 +119,3 @@
 
         st.write("### Treatment")
         st.dataframe(data=pd.DataFrame.from_dict(st.session_state.treatment_stats))
-

pages/2_Analyze_with_Summary_CSV.py

@@ -98,7 +98,14 @@
             options=[
                 c
                 for c in df.columns
-                if c not in [label_values, "sales", "exposure_period", "exposures", "conversions"]
+                if c
+                not in [
+                    label_values,
+                    "sales",
+                    "exposure_period",
+                    "exposures",
+                    "conversions",
+                ]
             ],
             help="Select which column refers to the value that comes from conversions, e.g.: revenue or cost from conversons.",
         )

requirements.txt

@@ -1,4 +1,4 @@
-pandas==1.2.5
+pandas==1.4.4
 numpy==1.25.1
 scipy==1.11.1
 streamlit==1.13.0

setup.py

@@ -8,7 +8,7 @@
     author="Luan Fernandes",
     license="",
     install_requires=[
-        "pandas==1.2.5",
+        "pandas==1.4.4",
         "numpy==1.25.1",
         "scipy==1.11.1",
     ],

tests/test_experiment.py

@@ -23,7 +23,7 @@ def test_run_experiment_one_variant(self):
         }
         experiment = Experiment(**args)
 
-        df_results_per_user = pd.read_csv("tests/results_per_user.csv")
+        df_results_per_user = pd.read_csv("examples/results_per_user.csv")
         with self.assertRaises(AssertionError):
             experiment.run_experiment(df_results_per_user=df_results_per_user)
 

tests/test_variants.py

@@ -9,8 +9,7 @@ def test_init(self):
         self.assertEqual(variant.name, "A")
 
     def test_consolidate_results(self):
-
-        results_per_user = pd.read_csv("tests/results_per_user.csv")
+        results_per_user = pd.read_csv("examples/results_per_user.csv")
         variant_results = {
             "A": {
                 "users": 6,

utils.py

@@ -11,10 +11,13 @@ def save_results_in_session_state(experiment, control_label, treatment_label):
         experiment.results[control_label]["statistics"]
     )
 
+
 def explain_metrics():
-    st.write("""
+    st.write(
+        """
     - **chance_to_beat**: chance that the variant is better than the other.
     - **expected_loss**: a measure of the risk you're assuming if you stay with this variant. The lower the risk, the best (e.g: 0.10 in conversion means that your risk of staying with the variant compared to the other is to lose 10% p.p. 0.10 for arpu is a risk of loosing $0.10 per user.)
     - **lift**: the observed relative difference compared to the other variant for each metric.
     - **diff**: the observed absolute difference compared to the other variant for each metric.
-    """)
+    """
+    )