import streamlit as st import joblib import pandas as pd # st.title('Placement Prediction app') # st.markdown(""" # # """, unsafe_allow_html=True) # st.title('Placement Prediction App') # st.subheader('Predicting student placement outcomes using machine learning') # st.markdown('This app uses historical data to predict whether a student will be placed in a company based on their profile.') try: model = joblib.load('model_campus') st.success("Model loaded successfully!") except Exception as e: st.error(f"Error loading model: {e}") st.stop() # model = joblib.load(open('model_campus)) def predict_placement(data): # Preprocess the data new_data = pd.DataFrame(data, index=[0]) # Make prediction prediction = model.predict(new_data)[0] prob = model.predict_proba(new_data)[0][1] return prediction, prob def main(): st.header('Placement Prediciton App') gender = st.radio('Gender', ['Male', 'Female']) ssc_p = st.number_input('Secondary School Percentage', min_value=0.0, max_value=100.0, value=50.0, step=0.1) ssc_b = st.radio('Board of Education (SSC)', ['Central', 'Others']) hsc_p = st.number_input('Higher Secondary Percentage', min_value=0.0, max_value=100.0, value=50.0, step=0.1) hsc_b = st.radio('Board of Education (HSC)', ['Central', 'Others']) degree_p = st.number_input('UG Percentage', min_value=0.0, max_value=100.0, value=50.0, step=0.1) branch = st.selectbox('Branch of Study', ['CSE', 'ECE/EN', 'Others']) workex = st.radio('Work Experience', ['Yes', 'No']) certifications = st.number_input('Number of Certifications', min_value=0, max_value=10, value=0) etest_p = st.number_input('Employability Test Score', min_value=0.0, max_value=100.0, value=50.0, step=0.1) backlogs = st.number_input('Number of Backlogs', min_value=0, max_value=10, value=0) if st.button('predict'): new_data = { 'gender': 0 if gender == "Male" else 1, 'ssc_p': ssc_p, 'ssc_b': 1 if ssc_b == "Central" else 0, 'hsc_p': hsc_p, 'hsc_b': 1 if hsc_b == "Central" else 0, 'degree_p': degree_p, 'Branch': 2 if branch == "ECE/EN" else 1 if branch == "CSE" else 0, 'Workex': 1 if workex == "Yes" else 0, 'Certifications': certifications, 'etest_p': etest_p, 'Backlogs': backlogs, } prediction, probability = predict_placement(new_data) st.write(f'Percentage of getting placed: {probability*100:.2f}%') if __name__=='__main__': main()