Track model file with Git LFS

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.keras filter=lfs diff=lfs merge=lfs -text

Weights/dropout_predictor.h5 → Models/model.keras

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d930d79a4c867d3ef823fa6c212c7e9faee5472b2410ac8f6c26e3e3f9eaeb7c
-size 213000
+oid sha256:2215cea53fda3f6f615f50c4748b2ab7bcc758807cd980f2378d532877213d2b
+size 220300

app.py

@@ -3,111 +3,69 @@ import numpy as np
 import tensorflow as tf
 import pickle
 
-from tensorflow.keras import layers, models
-
 app = Flask(__name__)
 
-model = models.Sequential([
-    layers.Input(shape=(9,)),
-    layers.Dense(128, activation='relu'),
-    layers.BatchNormalization(),
-    layers.Dropout(0.3),
-    layers.Dense(64, activation='relu'),
-    layers.BatchNormalization(),
-    layers.Dropout(0.2),
-    layers.Dense(32, activation='relu'),
-    layers.Dense(1, activation='sigmoid')  # Output: probability of dropout
-])
-
-model.load_weights(r"Weights/dropout_predictor.h5")
-
+# Load model and scaler
+model = tf.keras.models.load_model(r"Models/model.keras")
 with open("scaler.pkl", "rb") as f:
     scaler = pickle.load(f)
 
-gender_map = {"Male": 0, "Female": 1}
-extracurricular_map = {"No": 0, "Yes": 1}
+# LabelEncoder mapping (as used during training)
+gender_map = {"Male": 1, "Female": 0}
+extra_map = {"Yes": 1, "No": 0}
 
-expected_features = [
-    "Current_Sem", "Gender", "Extracurricular",
-    "SGPA_Sem_1", "SGPA_Sem_2", "SGPA_Sem_3", "SGPA_Sem_4", "SGPA_Sem_5", "SGPA_Sem_6",
-    "Attendance_Sem_1", "Attendance_Sem_2", "Attendance_Sem_3",
-    "Attendance_Sem_4", "Attendance_Sem_5", "Attendance_Sem_6"
-]
+# Expected input fields
+semesters = range(1, 7)
+base_features = ["Current_Sem", "Gender", "Extracurricular"]
+sgpa_fields = [f"SGPA_Sem_{i}" for i in semesters]
+att_fields = [f"Attendance_Sem_{i}" for i in semesters]
+mask_fields = [f"{col}_mask" for col in sgpa_fields + att_fields]
 
-from flask import request, jsonify
-import numpy as np
+input_order = base_features + sgpa_fields + att_fields + mask_fields
 
 @app.route("/predict", methods=["POST"])
 def predict():
     try:
         data = request.get_json()
-
         if not isinstance(data, list):
-            return jsonify({"error": "Expected a list of student records"}), 400
-
-        response_data = []
-
-        for student in data:
-            try:
-                name = student["name"]
-                usn = student["usn"]
-                gender_str = student["gender"]
-                extra_raw = student.get("extracurricular", 0)
-                current_sem = student["current_sem"]
-                semesters = student["semesters"]
-
-                # Normalize gender
-                gender = gender_map.get(gender_str.lower(), 0)
-
-                # Normalize extracurricular (out of 10)
-                extra = float(extra_raw) / 10 if extra_raw else 0.0
-
-                # Aggregate SGPA and attendance per semester
-                sgpas = []
-                atts = []
-
-                for sem in range(1, 7):  # Semesters 1 to 6
-                    sem_data = next((s for s in semesters if s["semester_number"] == sem), None)
-
-                    if sem_data:
-                        subjects = sem_data.get("subjects", [])
-                        marks = [sub["marks"] / 100 for sub in subjects if "marks" in sub]
-                        attendance = [sub["attendance"] / 100 for sub in subjects if "attendance" in sub]
-
-                        sgpa = np.mean(marks) if marks else 0.0
-                        att = np.mean(attendance) if attendance else 0.0
-                    else:
-                        sgpa = 0.0
-                        att = 0.0
-
-                    sgpas.append(sgpa)
-                    atts.append(att)
-
-                # Build feature vector
-                features = [current_sem, gender, extra] + sgpas + atts
-                X = scaler.transform([features])
-                prob = float(model.predict(X).flatten()[0])
-
-                # Enrich student JSON
-                enriched_student = {
-                    **student,
-                    "sgpa_per_semester": sgpas,
-                    "attendance_per_semester": atts,
-                    "risk": prob
-                }
-
-                response_data.append(enriched_student)
-
-            except Exception as student_error:
-                response_data.append({
-                    "error": f"Error processing student {student.get('usn', 'unknown')}: {str(student_error)}"
-                })
-
-        return jsonify(response_data)
+            return jsonify({"error": "Expected a list of input records"}), 400
+
+        processed = []
+        for record in data:
+            # Validate required base fields
+            for field in base_features:
+                if field not in record:
+                    return jsonify({"error": f"Missing required field: {field}"}), 400
+
+            # Encode categorical fields
+            gender = gender_map.get(record["Gender"], 0)
+            extra = extra_map.get(record["Extracurricular"], 0)
+
+            # Handle SGPA and Attendance with masks
+            sgpas = []
+            s_masks = []
+            for field in sgpa_fields:
+                val = record.get(field, None)
+                sgpas.append(float(val) if val is not None else 0.0)
+                s_masks.append(1 if val is not None else 0)
+
+            atts = []
+            a_masks = []
+            for field in att_fields:
+                val = record.get(field, None)
+                atts.append(float(val) if val is not None else 0.0)
+                a_masks.append(1 if val is not None else 0)
+
+            row = [record["Current_Sem"], gender, extra] + sgpas + atts + s_masks + a_masks
+            processed.append(row)
+
+        # Scale inputs
+        X = scaler.transform(np.array(processed))
+        probs = model.predict(X).flatten()
+        return jsonify([{"dropout_probability": float(p)} for p in probs])
 
     except Exception as e:
         return jsonify({"error": str(e)}), 500
 
-
 if __name__ == "__main__":
     app.run(debug=True)

scaler.pkl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a72d56371da73a9575f07e125500f4bec157184e5c92fd880222b513ebd3705b
+oid sha256:2db951be4c8d024a6f824113ce0e7951919eea9a79c6bfdba31c31c3c7a651bf
 size 1247