Update app.py

app.py

@@ -1,105 +1,43 @@
-import streamlit as st
+
+   # app.py
+import gradio as gr
 import pandas as pd
 import numpy as np
-from sklearn.preprocessing import LabelEncoder
-from sklearn.tree import DecisionTreeClassifier
-from sklearn.neighbors import KNeighborsClassifier
-from sklearn.metrics import accuracy_score
-import matplotlib.pyplot as plt
-import seaborn as sns
-from reportlab.pdfgen import canvas
-from io import BytesIO
-
-# Load dataset
-data = pd.read_csv("students.csv")
-
-# Encode categorical data
-le_interest = LabelEncoder()
-le_course = LabelEncoder()
-data["interest"] = le_interest.fit_transform(data["interest"])
-data["recommended_course"] = le_course.fit_transform(data["recommended_course"])
-
-# Features and target
-X = data[["math_grade","programming_grade","gpa","interest"]]
-y = data["recommended_course"]
-
-# Train Hybrid Models
-dt_model = DecisionTreeClassifier()
-dt_model.fit(X, y)
-knn_model = KNeighborsClassifier(n_neighbors=3)
-knn_model.fit(X, y)
-
-# Hybrid prediction function
-def recommend_courses(math, programming, gpa, interest_text, top_n=3):
-    interest_encoded = le_interest.transform([interest_text])[0]
-    X_new = np.array([[math, programming, gpa, interest_encoded]])
+from sklearn.metrics.pairwise import cosine_similarity
+from sklearn.preprocessing import StandardScaler
+
+# 1️⃣ Data ya courses (urugero)
+data = pd.DataFrame({
+    'Course': ['Mathematics', 'Physics', 'Computer Science', 'Chemistry', 'Biology'],
+    'Difficulty': [3, 4, 2, 4, 3],
+    'Popularity': [5, 4, 5, 3, 4]
+})
+
+# 2️⃣ Preprocess
+scaler = StandardScaler()
+features = scaler.fit_transform(data[['Difficulty', 'Popularity']])
+
+# 3️⃣ Recommendation function
+def recommend(course_name, top_n=3):
+    if course_name not in data['Course'].values:
+        return "Course not found!"
     
-    dt_pred = dt_model.predict_proba(X_new)[0]
-    knn_pred = knn_model.predict_proba(X_new)[0]
+    idx = data.index[data['Course'] == course_name][0]
+    sim_scores = cosine_similarity([features[idx]], features)[0]
     
-    avg_pred = (dt_pred + knn_pred)/2
-    top_indices = avg_pred.argsort()[-top_n:][::-1]
-    courses = le_course.inverse_transform(top_indices)
-    return courses
-
-# Streamlit UI
-st.set_page_config(page_title="Advanced Course Recommender", layout="wide")
-st.title("🎓 Advanced University Course Recommendation System")
-
-st.sidebar.header("Input Student Data")
-math = st.sidebar.slider("Math Grade",0,100,70)
-programming = st.sidebar.slider("Programming Grade",0,100,70)
-gpa = st.sidebar.slider("GPA",0.0,4.0,3.0,0.1)
-interest = st.sidebar.selectbox("Interest", le_interest.classes_)
-
-if st.button("Recommend Courses"):
-    top_courses = recommend_courses(math, programming, gpa, interest, top_n=3)
-    st.success(f"✅ Top 3 Recommended Courses: {', '.join(top_courses)}")
-    
-    # Show student input
-    st.subheader("📊 Your Input Data")
-    st.table(pd.DataFrame({
-        "Math Grade":[math],
-        "Programming Grade":[programming],
-        "GPA":[gpa],
-        "Interest":[interest]
-    }))
+    recommended_idx = np.argsort(sim_scores)[::-1][1:top_n+1]
+    recommended_courses = data['Course'].iloc[recommended_idx].tolist()
     
-    # Graph: Math vs Programming comparison
-    st.subheader("📈 Math vs Programming Comparison")
-    fig, ax = plt.subplots(figsize=(8,4))
-    sns.scatterplot(data=data, x="math_grade", y="programming_grade",
-                    hue="recommended_course", palette="tab10", s=100, ax=ax)
-    ax.scatter(math, programming, color='red', s=200, label="You")
-    ax.set_title("Math vs Programming Grades")
-    ax.legend()
-    st.pyplot(fig)
-    
-    # PDF download
-    st.subheader("📄 Download Recommendation PDF")
-    buffer = BytesIO()
-    c = canvas.Canvas(buffer, pagesize=(400, 400))
-    c.setFont("Helvetica-Bold", 14)
-    c.drawString(50,350,"Course Recommendation Report")
-    c.setFont("Helvetica", 12)
-    c.drawString(50,320,f"Math Grade: {math}")
-    c.drawString(50,300,f"Programming Grade: {programming}")
-    c.drawString(50,280,f"GPA: {gpa}")
-    c.drawString(50,260,f"Interest: {interest}")
-    c.drawString(50,240,"Recommended Courses:")
-    for i, course in enumerate(top_courses):
-        c.drawString(70,220-20*i,f"{i+1}. {course}")
-    c.showPage()
-    c.save()
-    buffer.seek(0)
-    st.download_button("Download PDF", buffer, file_name="recommendation_report.pdf", mime="application/pdf")
-
-with st.expander("View Dataset"):
-    st.dataframe(data)
-
-# Accuracy metrics
-st.subheader("🔍 Model Accuracy Metrics")
-dt_acc = accuracy_score(y, dt_model.predict(X))
-knn_acc = accuracy_score(y, knn_model.predict(X))
-st.write(f"Decision Tree Accuracy: {dt_acc*100:.2f}%")
-st.write(f"KNN Accuracy: {knn_acc*100:.2f}%")
+    return recommended_courses
+
+# 4️⃣ Gradio interface
+iface = gr.Interface(
+    fn=recommend,
+    inputs=[gr.Dropdown(choices=data['Course'].tolist(), label="Select a course"), 
+            gr.Slider(1, 5, step=1, label="Number of recommendations")],
+    outputs="text",
+    title="University Course Recommendation System",
+    description="Select a course and get similar course recommendations."
+)
+
+iface.launch()