import os import sys from random import randint import time import uuid import argparse sys.path.append(os.path.abspath("../supv")) from matumizi.util import * from mcclf import * import streamlit as st def genVisitHistory(numUsers, convRate, label): for i in range(numUsers): userID = genID(12) userSess = [] userSess.append(userID) conv = randint(0, 100) if (conv < convRate): #converted if (label): if (randint(0,100) < 90): userSess.append("T") else: userSess.append("F") numSession = randint(2, 20) for j in range(numSession): sess = randint(0, 100) if (sess <= 15): elapsed = "H" elif (sess > 15 and sess <= 40): elapsed = "M" else: elapsed = "L" sess = randint(0, 100) if (sess <= 15): duration = "L" elif (sess > 15 and sess <= 40): duration = "M" else: duration = "H" sessSummary = elapsed + duration userSess.append(sessSummary) else: #not converted if (label): if (randint(0,100) < 90): userSess.append("F") else: userSess.append("T") numSession = randint(2, 12) for j in range(numSession): sess = randint(0, 100) if (sess <= 20): elapsed = "L" elif (sess > 20 and sess <= 45): elapsed = "M" else: elapsed = "H" sess = randint(0, 100) if (sess <= 20): duration = "H" elif (sess > 20 and sess <= 45): duration = "M" else: duration = "L" sessSummary = elapsed + duration userSess.append(sessSummary) st.write(",".join(userSess)) def main(): st.set_page_config(page_title="Customer Conversion Prediction", page_icon=":guardsman:", layout="wide") st.title("Customer Conversion Prediction") # # Add sidebar # st.sidebar.title("Navigation") # app_mode = st.sidebar.selectbox("Choose the app mode", # ["Instructions", "Generate User Visit History", "Train Model", "Predict Conversion"]) # Add sidebar st.sidebar.title("Navigation") app_mode = st.sidebar.selectbox("Choose the App Mode", ["Instructions", "Generate User Visit History", "Predict Conversion"]) if app_mode == "Instructions": st.write("Welcome to the Markov Chain Classifier app!") # st.write("This app allows you to generate user visit history, train a Markov Chain Classifier model, and predict conversion.") st.write("This app allows you to generate user visit history and predict conversion of the visitor into customer.") st.write("To get started, use the sidebar to navigate to the desired functionality.") st.write("1. **Generate User Visit History**: Select the number of users and conversion rate, and click the 'Generate' button to generate user visit history.") # st.write("2. **Train Model**: Upload an ML config file using the file uploader, and click the 'Train' button to train the Markov Chain Classifier model.") st.write("2. **Predict Conversion**: Enter the User ID for which you want to predict, and click the 'Predict' button to make predictions whether the visitor will likely to convert into customer or not.") # Description of MarkovChainClassifier mcclf_description = "The MarkovChainClassifier is a Machine Learning Classifier that utilizes the concept of Markov chains for prediction. Markov chains are mathematical models that represent a system where the future state of the system depends only on its current state, and not on the previous states. The MarkovChainClassifier uses this concept to make predictions by modeling the transition probabilities between different states or categories in the input data. It captures the probabilistic relationships between variables and uses them to classify new data points into one or more predefined categories. The MarkovChainClassifier can be useful in scenarios where the data has a sequential or time-dependent structure, and the relationships between variables can be modeled as Markov chains. It can be applied to various tasks, such as text classification, speech recognition, recommendation systems, and financial forecasting." # Display the description in Streamlit app st.header("Model description:") st.write(mcclf_description) elif app_mode == "Generate User Visit History": st.subheader("Generate User Visit History") num_users = st.number_input("Number of users", min_value=1, max_value=10000, value=100, step=1) conv_rate = st.slider("Conversion rate", min_value=0, max_value=100, value=10, step=1) add_label = st.checkbox("Add label", value=False) if st.button("Generate"): genVisitHistory(num_users, conv_rate, add_label) # elif app_mode == "Train Model": # st.subheader("Train Model") # mlf_path = st.file_uploader("Upload ML config file") # if st.button("Train"): # if mlf_path is not None: # model = MarkovChainClassifier(mlf_path) # model.train() elif app_mode == "Predict Conversion": st.subheader("Predict Conversion") # Create an instance of MarkovChainClassifier with the ML config file model = MarkovChainClassifier("mcclf_cc.properties") # Get user input for userID user_id = st.text_input("Enter User ID") # Check if the "Predict" button was clicked if st.button("Predict"): # Call the predict method of the MarkovChainClassifier instance pred = model.predict() if pred == 'F': st.write(f"UserID: {user_id}, Prediction: Visitor is unlikely to convert into a customer.") else: st.write(f"UserID: {user_id}, Prediction: Visitor is likely to convert into a customer.") # st.subheader("Predict Conversion") # # Upload ML config file using Streamlit's file_uploader function # mlf_file = st.file_uploader("Upload ML config file", type=["properties"]) # # Check if ML config file was uploaded # if mlf_file is not None: # # Save the uploaded file to a local file # with open("mcclf_cc.properties", "wb") as f: # f.write(mlf_file.read()) # # Create an instance of MarkovChainClassifier with the uploaded ML config file # model = MarkovChainClassifier("mcclf_cc.properties") # # # Load the model from cc.mod # # model = MarkovChainClassifier.load_model("cc.mod") # # Get user input for userID # user_id = st.text_input("Enter User ID") # # Check if the "Predict" button was clicked # if st.button("Predict"): # # Load the saved model # # model.load_model("cc.mod") # # Call the predict method of the MarkovChainClassifier instance # pred = model.predict() # if pred == 'T': # st.write(f"UserID: {user_id}, Prediction: Visitor is likely to convert into a customer.") # else: # st.write(f"UserID: {user_id}, Prediction: Visitor is unlikely to convert into a customer.") if __name__ == "__main__": main()