Update app.py

app.py

@@ -9,6 +9,7 @@ import gradio as gr
 ######################################
 df = pd.read_csv("X_train_Y_Train_merged_train.csv")
 
+# List of model filenames (adjust if needed)
 model_filenames = [
     "YOWRCONC.pkl", "YOSEEDOC.pkl", "YO_MDEA5.pkl", "YOWRLSIN.pkl",
     "YODPPROB.pkl", "YOWRPROB.pkl", "YODPR2WK.pkl", "YOWRDEPR.pkl",
@@ -17,6 +18,7 @@ model_filenames = [
 ]
 model_path = "models/"
 
+
 ######################################
 # 2) Model Predictor
 ######################################
@@ -59,7 +61,10 @@ class ModelPredictor:
         return models
 
     def make_predictions(self, user_input):
-        # Each model => returns array of [0] or [1]
+        """
+        Returns a list of numpy arrays, each array is [0] or [1].
+        The i-th array corresponds to the i-th model in self.models.
+        """
         predictions = []
         for model in self.models:
             pred = model.predict(user_input)
@@ -67,13 +72,16 @@ class ModelPredictor:
         return predictions
 
     def get_majority_vote(self, predictions):
+        """
+        Flatten all predictions from all models, combine them,
+        then find the majority class (0 or 1).
+        """
         combined = np.concatenate(predictions)
-        # 0 or 1 with highest frequency
-        majority_vote = np.bincount(combined).argmax()
-        return majority_vote
+        majority = np.bincount(combined).argmax()
+        return majority
 
+    # Simple threshold approach (0-4 => Very Low, 5-8 => Low, etc.)
     def evaluate_severity(self, majority_vote_count):
-        # Simple threshold approach
         if majority_vote_count >= 13:
             return "Mental Health Severity: Severe"
         elif majority_vote_count >= 9:
@@ -83,6 +91,7 @@ class ModelPredictor:
         else:
             return "Mental Health Severity: Very Low"
 
+
 ######################################
 # 3) Validate Inputs
 ######################################
@@ -92,6 +101,7 @@ def validate_inputs(*args):
             return False
     return True
 
+
 ######################################
 # 4) Core Prediction
 ######################################
@@ -104,6 +114,26 @@ def predict(
     YHLTMDE, YDOCMDE, YTXMDEYR, YUSUITHKYR, YUSUIPLNYR, YUSUITHK, 
     YUSUIPLN, MDEIMPY, LVLDIFMEM2, YMSUD5YANY, YRXMDEYR
 ):
+    # Validate
+    if not validate_inputs(
+        YMDEYR, YMDERSUD5ANY, YMDEIMAD5YR, YMIMS5YANY, YMDELT, YMDEHARX,
+        YMDEHPRX, YMDETXRX, YMDEHPO, YMDEAUD5YR, YMIMI5YANY, YMIUD5YANY,
+        YMDESUD5ANYO, YNURSMDE, YSOCMDE, YCOUNMDE, YPSY1MDE, YPSY2MDE,
+        YHLTMDE, YDOCMDE, YTXMDEYR, YUSUITHKYR, YUSUIPLNYR, YUSUITHK,
+        YUSUIPLN, MDEIMPY, LVLDIFMEM2, YMSUD5YANY, YRXMDEYR
+    ):
+        return (
+            "Please select all required fields.",
+            "Validation Error",
+            "No data",
+            None,
+            "No data",
+            None,
+            None,
+            None
+        )
+
+    # Build dataframe from user inputs
     user_input_data = {
         'YNURSMDE': [int(YNURSMDE)],
         'YMDEYR': [int(YMDEYR)],
@@ -137,48 +167,46 @@ def predict(
     }
     user_input = pd.DataFrame(user_input_data)
 
-    # 1) Predict
+    # 1) Predictions
     predictions = predictor.make_predictions(user_input)
 
     # 2) Majority vote
     majority_vote = predictor.get_majority_vote(predictions)
 
-    # 3) Count how many are '1'
+    # 3) Count of '1's
     num_ones = sum(np.concatenate(predictions) == 1)
 
     # 4) Severity
     severity = predictor.evaluate_severity(num_ones)
 
-    # 5) Grouped textual results
-    #    [Same grouping logic as before, or adapt as needed]
+    # 5) Group textual results
     groups = {
         "Concentration_and_Decision_Making": ["YOWRCONC", "YOWRDCSN"],
         "Sleep_and_Energy_Levels": ["YOWRHRS", "YO_MDEA5", "YOWRELES", "YO_MDEA2"],
-        "Mood_and_Emotional_State": ["YOWRCHR", "YOWRLSIN", "YOWRDEPR", "YODPDISC", 
+        "Mood_and_Emotional_State": ["YOWRCHR", "YOWRLSIN", "YOWRDEPR", "YODPDISC",
                                      "YOLOSEV", "YODPLSIN", "YODSCEV"],
         "Appetite_and_Weight_Changes": ["YO_MDEA3", "YOWRELES"],
-        "Duration_and_Severity_of_Depression_Symptoms": ["YODPPROB", "YOWRPROB", 
-                                                         "YODPR2WK", "YODSMMDE", 
+        "Duration_and_Severity_of_Depression_Symptoms": ["YODPPROB", "YOWRPROB",
+                                                         "YODPR2WK", "YODSMMDE",
                                                          "YOPB2WK"]
     }
+
     grouped_text = {k: [] for k in groups}
-    for i, pred in enumerate(predictions):
+    for i, arr in enumerate(predictions):
         col_name = model_filenames[i].split('.')[0]
-        pred_val = pred[0]
+        pred_val = arr[0]
         if col_name in predictor.prediction_map and pred_val in [0,1]:
             text_val = predictor.prediction_map[col_name][pred_val]
         else:
             text_val = f"Prediction={pred_val}"
-        # Find which group
-        assigned = False
+
+        found_group = False
         for gname, gcols in groups.items():
             if col_name in gcols:
                 grouped_text[gname].append(f"{col_name} => {text_val}")
-                assigned = True
+                found_group = True
                 break
-        if not assigned:
-            # Or skip
-            pass
+        # If not found_group, we do nothing (skip or put in a "misc" group)
 
     final_str = []
     for gname, items in grouped_text.items():
@@ -190,7 +218,7 @@ def predict(
     if not final_str:
         final_str = "No predictions made. Please check inputs."
 
-    # 6) Additional charts: total patients, distribution for input features, etc.
+    # Additional info
     total_patients = len(df)
     total_patient_markdown = (
         f"### Total Patient Count\nThere are **{total_patients}** patients in the dataset."
@@ -203,18 +231,20 @@ def predict(
         same_val_counts[col] = len(df[df[col] == val_])
     bar_input_df = pd.DataFrame({"Feature": list(same_val_counts.keys()),
                                  "Count": list(same_val_counts.values())})
-    fig_bar_input = px.bar(bar_input_df, x="Feature", y="Count",
-                           title="Number of Patients with Same Input Feature Values")
+    fig_bar_input = px.bar(
+        bar_input_df, x="Feature", y="Count",
+        title="Number of Patients with Same Input Feature Values"
+    )
     fig_bar_input.update_layout(width=800, height=500)
 
     # B) Bar chart for predicted labels
     label_counts = {}
-    all_preds_flat = np.concatenate(predictions)
     for i, arr in enumerate(predictions):
         lbl_col = model_filenames[i].split('.')[0]
         pred_val = arr[0]
         if pred_val in [0,1]:
             label_counts[lbl_col] = len(df[df[lbl_col] == pred_val])
+
     if label_counts:
         bar_label_df = pd.DataFrame({"Label": list(label_counts.keys()),
                                      "Count": list(label_counts.values())})
@@ -226,9 +256,8 @@ def predict(
         fig_bar_labels.update_layout(width=800, height=500)
 
     # C) Distribution Plot (small sample)
-    #    We'll pick the first 4 user_input columns & first 3 labels
-    subset_input_cols = list(user_input_data.keys())[:4]
-    subset_labels = [fn.split('.')[0] for fn in model_filenames[:3]]
+    subset_input_cols = list(user_input_data.keys())[:4]  # first 4 input columns
+    subset_labels = [fn.split('.')[0] for fn in model_filenames[:3]]  # first 3 label columns
     dist_rows = []
     for feat in subset_input_cols:
         if feat not in df.columns:
@@ -242,37 +271,38 @@ def predict(
             dist_rows.append(tmp)
     if dist_rows:
         big_dist_df = pd.concat(dist_rows, ignore_index=True)
-        fig_dist = px.bar(big_dist_df,
-                          x=big_dist_df.columns[0],
-                          y="count",
-                          color=big_dist_df.columns[1],
-                          facet_row="feature",
-                          facet_col="label",
-                          title="Distribution of Sample Input Features vs. Sample Predicted Labels")
+        fig_dist = px.bar(
+            big_dist_df,
+            x=big_dist_df.columns[0],
+            y="count",
+            color=big_dist_df.columns[1],
+            facet_row="feature",
+            facet_col="label",
+            title="Distribution of Sample Input Features vs. Sample Predicted Labels"
+        )
         fig_dist.update_layout(width=1000, height=700)
     else:
         fig_dist = px.bar(title="Distribution plot not generated.")
 
-    # D) Nearest Neighbors (K=2) [Optional as before]
-    #    ... omitted for brevity if you want to keep from prior code ...
-    #    or keep it. 
-    # For now, let's produce an empty markdown
-    nearest_neighbors_markdown = "Nearest neighbors omitted here for brevity..."
+    # D) Nearest neighbors (placeholder or your own logic)
+    nearest_neighbors_markdown = "Nearest neighbors omitted or placed here if needed..."
 
-    # We won't produce a default co-occurrence plot here, since we do it in a separate tab.
+    # We won't produce a co-occurrence plot by default here, so set to None
+    co_occurrence_placeholder = None
 
-    # Return 8 items
+    # Return the 8 outputs
     return (
-        final_str, 
-        severity, 
-        total_patient_markdown,
-        fig_dist,
-        nearest_neighbors_markdown,
-        None,  # placeholder for a single co-occurrence plot
-        fig_bar_input,
-        fig_bar_labels
+        final_str,           # 1) Prediction Results
+        severity,            # 2) Mental Health Severity
+        total_patient_markdown,  # 3) Total Patient Count
+        fig_dist,            # 4) Distribution Plot
+        nearest_neighbors_markdown, # 5) Nearest Neighbors
+        co_occurrence_placeholder,  # 6) Co-occurrence Plot placeholder
+        fig_bar_input,       # 7) Bar Chart for input features
+        fig_bar_labels       # 8) Bar Chart for predicted labels
     )
 
+
 ######################################
 # 5) Input Mapping
 ######################################
@@ -308,20 +338,19 @@ input_mapping = {
     'YMDELT': {"Yes": 1, "No": 2}
 }
 
+
 ######################################
-# 6) Co-Occurrence Function (Separate)
+# 6) Co-Occurrence Function
 ######################################
 def co_occurrence_plot(feature1, feature2, label_col):
     """
     Generate a single co-occurrence bar chart grouping by [feature1, feature2, label_col].
-    We set a custom width/height so it's clearly visible.
     """
     if not feature1 or not feature2 or not label_col:
         return px.bar(title="Please select all three fields.")
     if feature1 not in df.columns or feature2 not in df.columns or label_col not in df.columns:
         return px.bar(title="Selected columns not found in the dataset.")
 
-    # Group
     grouped_df = df.groupby([feature1, feature2, label_col]).size().reset_index(name="count")
     fig = px.bar(
         grouped_df,
@@ -334,13 +363,14 @@ def co_occurrence_plot(feature1, feature2, label_col):
     fig.update_layout(width=1000, height=600)
     return fig
 
+
 ######################################
-# 7) Gradio with Tabs
+# 7) Gradio Interface with Tabs
 ######################################
 with gr.Blocks(css=".gradio-container {max-width: 1200px;}") as demo:
 
     with gr.Tab("Prediction"):
-        # Inputs (same order as function signature)
+        # --------- INPUT FIELDS --------- #
         YMDEYR_dd = gr.Dropdown(list(input_mapping['YMDEYR'].keys()), label="YMDEYR")
         YMDERSUD5ANY_dd = gr.Dropdown(list(input_mapping['YMDERSUD5ANY'].keys()), label="YMDERSUD5ANY")
         YMDEIMAD5YR_dd = gr.Dropdown(list(input_mapping['YMDEIMAD5YR'].keys()), label="YMDEIMAD5YR")
@@ -377,7 +407,10 @@ with gr.Blocks(css=".gradio-container {max-width: 1200px;}") as demo:
         YMSUD5YANY_dd = gr.Dropdown(list(input_mapping['YMSUD5YANY'].keys()), label="YMSUD5YANY")
         YRXMDEYR_dd = gr.Dropdown(list(input_mapping['YRXMDEYR'].keys()), label="YRXMDEYR")
 
-        # 8 outputs
+        # --------- PREDICT BUTTON (BEFORE OUTPUTS) --------- #
+        predict_btn = gr.Button("Predict")
+
+        # --------- OUTPUTS (IN THE SAME ORDER AS THE RETURN TUPLE) --------- #
         out_pred_res = gr.Textbox(label="Prediction Results", lines=8)
         out_sev = gr.Textbox(label="Mental Health Severity", lines=2)
         out_count = gr.Markdown(label="Total Patient Count")
@@ -387,9 +420,6 @@ with gr.Blocks(css=".gradio-container {max-width: 1200px;}") as demo:
         out_bar_input = gr.Plot(label="Input Feature Counts")
         out_bar_labels = gr.Plot(label="Predicted Label Counts")
 
-        # Button
-        predict_btn = gr.Button("Predict")
-
         # Link button to the function
         predict_btn.click(
             fn=predict,
@@ -401,10 +431,12 @@ with gr.Blocks(css=".gradio-container {max-width: 1200px;}") as demo:
                 YUSUIPLN_dd, MDEIMPY_dd, LVLDIFMEM2_dd, YMSUD5YANY_dd, YRXMDEYR_dd
             ],
             outputs=[
-                out_pred_res, out_sev, out_count, out_distplot, out_nn, out_cooc, out_bar_input, out_bar_labels
+                out_pred_res, out_sev, out_count, out_distplot,
+                out_nn, out_cooc, out_bar_input, out_bar_labels
             ]
         )
 
+    # ------------- SECOND TAB (CO-OCCURRENCE) -------------
     with gr.Tab("Co-occurrence"):
         gr.Markdown("## Generate a Co-Occurrence Plot on Demand\nSelect two features and one label:")
         with gr.Row():
@@ -414,22 +446,11 @@ with gr.Blocks(css=".gradio-container {max-width: 1200px;}") as demo:
         out_co_occ_plot = gr.Plot(label="Co-occurrence Plot")
 
         co_occ_btn = gr.Button("Generate Plot")
-
-        # Link to co_occurrence_plot function
         co_occ_btn.click(
             fn=co_occurrence_plot,
             inputs=[feature1_dd, feature2_dd, label_dd],
             outputs=out_co_occ_plot
         )
 
-# Optional custom CSS for bigger container
-custom_css = """
-    .gradio-container {
-        max-width: 1200px;
-        margin-left: auto;
-        margin-right: auto;
-    }
-"""
-
-# Launch
+# Optionally, you can customize your CSS or server launch parameters
 demo.launch(server_name="0.0.0.0", server_port=7860)