app.py

import pandas as pd
from surprise import Dataset, Reader

laptop_df = pd.read_csv('laptop_data.csv')
user_df = pd.read_csv('user_data.csv')  

laptop_df = laptop_df.fillna(0)
user_df = user_df.fillna(0)

# Create a Surprise Dataset
reader = Reader(rating_scale=(0, 5))
data = Dataset.load_from_df(user_df[['User_ID', 'Laptop_ID', 'Rating']], reader)

from surprise.model_selection import train_test_split
from surprise import SVD
from surprise import accuracy

# Train-test split
trainset, testset = train_test_split(data, test_size=0.2, random_state=42)

# train model
model = SVD()
model.fit(trainset)

def recommend_laptops(age=None, category=None, gender=None, user_id=None, num_recommendations=5):
    if user_id is not None:
        # Existing user
        user_ratings = user_df[user_df['User_ID'] == user_id]
        user_unrated_laptops = laptop_df[~laptop_df['Laptop_ID'].isin(user_ratings['Laptop_ID'])]
        user_unrated_laptops['Predicted_Rating'] = user_unrated_laptops['Laptop_ID'].apply(lambda x: model.predict(user_id, x).est)
        recommendations = user_unrated_laptops.sort_values(by='Predicted_Rating', ascending=False).head(num_recommendations)
    else:
        # New user
        new_user_data = pd.DataFrame({
            'User_ID': [10002],  
            'Age': [age],
            'Category': [category],
            'Gender': [gender]
        })
        new_user_data = new_user_data.merge(laptop_df, how='cross')
        new_user_data['Predicted_Rating'] = new_user_data.apply(lambda row: model.predict(10002, row['Laptop_ID']).est, axis=1)
        recommendations = new_user_data.sort_values(by='Predicted_Rating', ascending=False).head(num_recommendations)

    return recommendations


import streamlit as st

# Streamlit app
st.title("Laptop Recommendation System")

# User choice: New or Existing user
user_type = st.radio("Are you a new user or an existing user?", ('New User', 'Existing User'))

if user_type == 'New User':
    # User input for new users
    new_user_age = st.slider("Age:", min_value=12, max_value=89, value=25)
    new_user_category = st.selectbox("What best describes you:", ['Student', 'Professor', 'Banker', 'Businessman', 'Programmer', 'Other'])
    new_user_gender = st.radio("Gender:", ['Male', 'Female', 'Other'])

# Button to get recommendations for new users
    if st.button("Get Recommendations"):
        recommendations = recommend_laptops(age=new_user_age, category=new_user_category, gender=new_user_gender)
        st.subheader("Top 5 Recommended Laptops:")

        # decoding features
        type_mapping = {1: 'gaming laptop', 2: 'thin and light laptop', 3: '2 in 1 laptop', 4: 'notebook', 5: 'laptop',
                6: '2 in 1 gaming laptop', 7: 'business laptop', 8: 'chromebook', 9: 'creator laptop'}

        processor_brand_mapping = {1: 'intel', 2: 'amd', 3: 'qualcomm', 4: 'apple', 5: 'mediatek'}
        
        os_mapping = {1: 'windows', 2: 'chrome os', 3: 'dos', 4: 'mac', 5: 'ubuntu'}
        
        company_mapping = {1: 'asus', 2: 'hp', 3: 'lenovo', 4: 'dell', 5: 'msi', 6: 'realme', 7: 'avita', 8: 'acer',
                           9: 'samsung', 10: 'infinix', 11: 'lg', 12: 'apple', 13: 'nokia', 14: 'redmibook',
                           15: 'mi', 16: 'vaio'}
        
        # Decode the encoded features
        recommendations['Type'] = recommendations['Type'].map(type_mapping)
        recommendations['Processor Brand'] = recommendations['Processor Brand'].map(processor_brand_mapping)
        recommendations['Operating System'] = recommendations['Operating System'].map(os_mapping)
        recommendations['company'] = recommendations['company'].map(company_mapping)

        boolean_columns = ['SSD', 'Expandable Memory', 'Touchscreen']
        for column in boolean_columns:
            recommendations[column] = recommendations[column].map({0: 'No', 1: 'Yes'})


        recommendations_table = recommendations[['name', 'Price (in Indian Rupees)', 'Type', 'Dedicated Graphic Memory Capacity',
                                      'Processor Brand', 'SSD', 'RAM (in GB)', 'RAM Type', 'Expandable Memory',
                                      'Operating System', 'Touchscreen', 'Screen Size (in inch)', 'Weight (in kg)',
                                      'Refresh Rate', 'screen_resolution', 'company', 'Storage', 'Processor name',
                                      'CPU_ranking', 'battery_backup', 'gpu name ', 'gpu_benchmark']]
    
        
        recommendations_table = recommendations_table.reset_index(drop=True)
        st.dataframe(recommendations_table)


# User input for existing users
elif user_type == 'Existing User':
    # User input for existing users
    existing_user_id = st.text_input("Enter your user ID:", "")

    # Button to get recommendations
    if st.button("Get Laptop Recommendations"):
        if existing_user_id:
            recommendations = recommend_laptops(user_id=int(existing_user_id))
            st.subheader(f"Top 5 Recommended Laptops for User {existing_user_id}:")
            # decoding features
            type_mapping = {1: 'gaming laptop', 2: 'thin and light laptop', 3: '2 in 1 laptop', 4: 'notebook', 5: 'laptop',
                    6: '2 in 1 gaming laptop', 7: 'business laptop', 8: 'chromebook', 9: 'creator laptop'}
    
            processor_brand_mapping = {1: 'intel', 2: 'amd', 3: 'qualcomm', 4: 'apple', 5: 'mediatek'}
            
            os_mapping = {1: 'windows', 2: 'chrome os', 3: 'dos', 4: 'mac', 5: 'ubuntu'}
            
            company_mapping = {1: 'asus', 2: 'hp', 3: 'lenovo', 4: 'dell', 5: 'msi', 6: 'realme', 7: 'avita', 8: 'acer',
                               9: 'samsung', 10: 'infinix', 11: 'lg', 12: 'apple', 13: 'nokia', 14: 'redmibook',
                               15: 'mi', 16: 'vaio'}
            
            # Decode the encoded features
            recommendations['Type'] = recommendations['Type'].map(type_mapping)
            recommendations['Processor Brand'] = recommendations['Processor Brand'].map(processor_brand_mapping)
            recommendations['Operating System'] = recommendations['Operating System'].map(os_mapping)
            recommendations['company'] = recommendations['company'].map(company_mapping)
    
            boolean_columns = ['SSD', 'Expandable Memory', 'Touchscreen']
            for column in boolean_columns:
                recommendations[column] = recommendations[column].map({0: 'No', 1: 'Yes'})
                
            recommendations_table = recommendations[['name', 'Price (in Indian Rupees)', 'Type', 'Dedicated Graphic Memory Capacity',
                                      'Processor Brand', 'SSD', 'RAM (in GB)', 'RAM Type', 'Expandable Memory',
                                      'Operating System', 'Touchscreen', 'Screen Size (in inch)', 'Weight (in kg)',
                                      'Refresh Rate', 'screen_resolution', 'company', 'Storage', 'Processor name',
                                      'CPU_ranking', 'battery_backup', 'gpu name ', 'gpu_benchmark']]

            recommendations_table = recommendations_table.reset_index(drop=True)
            st.dataframe(recommendations_table)
        else:
            st.warning("Please enter a valid user ID.")