Change UX to display all grades and graders

app.py

@@ -7,6 +7,7 @@ from statsmodels.stats.diagnostic import acorr_ljungbox
 import pickle
 import joblib
 import matplotlib.pyplot as plt
+from itertools import product
 
 # --- MongoDB Setup ---
 uri = "mongodb+srv://csmith715:I3xSO3ImRKFyQ0hf@cluster0.hc5mw.mongodb.net/"
@@ -87,6 +88,24 @@ def fetch_cert_data(certnumber):
     return df
 
 
+def transform_data(df):
+    df['ds'] = pd.to_datetime(df['ds'])
+    df['day_since'] = (pd.Timestamp.today().normalize() - df['ds']).dt.days
+    df['year'] = df['ds'].dt.year
+    df['month'] = df['ds'].dt.month
+    df['day_of_week'] = df['ds'].dt.dayofweek
+    df.drop('ds', axis=1, inplace=True)
+
+    df = pd.get_dummies(df, columns=['grader'])
+    df['grade'] = pd.to_numeric(df['grade'], errors='coerce')
+
+    poly = PolynomialFeatures(degree=3, include_bias=False)
+    poly_features = poly.fit_transform(df[['grade']])
+    poly_df = pd.DataFrame(poly_features, columns=['grade1', 'grade^2', 'grade^3'])
+    df = pd.concat([df, poly_df], axis=1).drop(columns=['grade1'])
+    return df
+
+
 class PokemonCardPredictor:
     def __init__(self):
         self.confidence_features = [
@@ -95,6 +114,7 @@ class PokemonCardPredictor:
             'reliability', 'day_since'
         ]
         self.latest_prices_df = pd.DataFrame()
+        self.full_df = pd.DataFrame()
 
     def plot_time_series(self, range_option):
         if self.latest_prices_df.empty:
@@ -137,31 +157,29 @@ class PokemonCardPredictor:
         plt.tight_layout()
         return fig
 
-
-    def transform_data(self, df):
-        df['ds'] = pd.to_datetime(df['ds'])
-        df['day_since'] = (pd.Timestamp.today().normalize() - df['ds']).dt.days
-        df['year'] = df['ds'].dt.year
-        df['month'] = df['ds'].dt.month
-        df['day_of_week'] = df['ds'].dt.dayofweek
-        df.drop('ds', axis=1, inplace=True)
-
-        df = pd.get_dummies(df, columns=['grader'])
-        df['grade'] = pd.to_numeric(df['grade'], errors='coerce')
-
-        poly = PolynomialFeatures(degree=3, include_bias=False)
-        poly_features = poly.fit_transform(df[['grade']])
-        poly_df = pd.DataFrame(poly_features, columns=['grade1', 'grade^2', 'grade^3'])
-        df = pd.concat([df, poly_df], axis=1).drop(columns=['grade1'])
-        return df
-
-    def predict(self, certnumber, grader, grade):
+    def predict_all(self, certnumber, grader, grade):
         raw_df = fetch_cert_data(certnumber)
         if raw_df.empty:
             self.latest_prices_df = pd.DataFrame()  # Reset
             return "Card info not found.", pd.DataFrame()
+        known_grades = raw_df['grade'].unique()
+        known_graders = raw_df['grader'].unique()
+
+        for k_grader, k_grade in product(known_graders, known_grades):
+            _, pred_df = self.predict(raw_df, k_grader, k_grade)
+            self.full_df = pd.concat([self.full_df, pred_df])
+        # Predict selected grade and grader for specific predictive purpose
+        pred, _ = self.predict(raw_df, grader, grade)
+
+        return f"Predicted Price: ${pred:,.2f}", self.full_df.round(2)
 
-        df = raw_df[(raw_df['grader'] == grader) & (raw_df['grade'] == grade)]
+    def predict(self, cert_df, grader, grade):
+        # raw_df = fetch_cert_data(certnumber)
+        # if raw_df.empty:
+        #     self.latest_prices_df = pd.DataFrame()  # Reset
+        #     return "Card info not found.", pd.DataFrame()
+
+        df = cert_df[(cert_df['grader'] == grader) & (cert_df['grade'] == grade)]
         if df.empty:
             self.latest_prices_df = pd.DataFrame()
             return "No transactions for this grader and grade.", pd.DataFrame()
@@ -180,7 +198,7 @@ class PokemonCardPredictor:
             return "No recent transaction to use.", pd.DataFrame()
 
         reliability = calculate_reliability(df)
-        transformed_df = self.transform_data(latest_df).fillna(0)
+        transformed_df = transform_data(latest_df).fillna(0)
         transformed_df = transformed_df[transformed_df['grade'] != 0]
 
         for col in gradient_boosting_model.feature_names_in_:
@@ -190,26 +208,36 @@ class PokemonCardPredictor:
         confidence_df = transformed_df.copy()
         confidence_df['reliability'] = reliability
         confidence_df['day_since'] = latest_df['day_since'].values
+        confidence_df = confidence_df[self.confidence_features].fillna(0)
+
+        risk_score = confidence_model.predict(confidence_df)
 
-        risk_score = confidence_model.predict(confidence_df[self.confidence_features])
-        prediction = gradient_boosting_model.predict(transformed_df[gradient_boosting_model.feature_names_in_])
+        transformed_df = transformed_df[gradient_boosting_model.feature_names_in_]
+        if transformed_df.empty:
+            return 'no data', pd.DataFrame()
+        prediction = gradient_boosting_model.predict(transformed_df)
 
         display_df = pd.DataFrame({
-            'Card Year': latest_df['card_year'].values,
+            'Grader': latest_df['grader'].values,
+            'Grade': latest_df['grade'].values,
+            # 'Card Year': latest_df['card_year'].values,
             'Name': latest_df['name'].values,
             'Set Name': latest_df['set_name'].values,
-            'Details': latest_df['details'].values,
+            # 'Details': latest_df['details'].values,
             'Predicted Price': prediction,
             'Risk': risk_score,
-            'ma_3d': latest_df['ma_3d'].values,
-            'ma_7d': latest_df['ma_7d'].values,
-            'ma_30d': latest_df['ma_30d'].values,
-            'count_3d': latest_df['count_3d'].values,
-            'count_7d': latest_df['count_7d'].values,
-            'count_30d': latest_df['count_30d'].values
+            'Most Recent Price': latest_df['y'].values,
+            'Days Since': latest_df['day_since'].values
+            # 'ma_3d': latest_df['ma_3d'].values,
+            # 'ma_7d': latest_df['ma_7d'].values,
+            # 'ma_30d': latest_df['ma_30d'].values,
+            # 'count_3d': latest_df['count_3d'].values,
+            # 'count_7d': latest_df['count_7d'].values,
+            # 'count_30d': latest_df['count_30d'].values
         })
 
-        return f"Predicted Price: ${prediction[0]:,.2f}", display_df.round(2)
+        # return f"Predicted Price: ${prediction[0]:,.2f}", display_df.round(2)
+        return prediction[0], display_df
 
 
 # --- Gradio UI ---
@@ -234,7 +262,7 @@ with gr.Blocks() as demo:
     output_plot = gr.Plot(label="Price Over Time")
 
     predict_btn.click(
-        fn=predictor.predict,
+        fn=predictor.predict_all,
         inputs=[cert_input, grader_input, grade_input],
         outputs=[output_text, output_table]
     ).then(