import streamlit as st import numpy as np from scipy import stats import matplotlib.pyplot as plt # Streamlit App Title st.title("📊 Z-Test for Website Conversion Rate Comparison") st.write("This app compares the conversion rate of an **old vs new website design** using a two-sample Z-test.") # --- Input Section --- st.header("🔢 Input Conversion Data") col1, col2 = st.columns(2) with col1: n1 = st.number_input("Visitors (Old Design):", min_value=1, value=10000) x1 = st.number_input("Conversions (Old Design):", min_value=0, value=500) with col2: n2 = st.number_input("Visitors (New Design):", min_value=1, value=10000) x2 = st.number_input("Conversions (New Design):", min_value=0, value=650) # Button to Run Z-Test if st.button("Run Z-Test"): # Calculate conversion rates p1 = x1 / n1 p2 = x2 / n2 # Pooled proportion p = (x1 + x2) / (n1 + n2) # Z-statistic z = (p2 - p1) / np.sqrt(p * (1 - p) * (1/n1 + 1/n2)) # p-value (one-tailed) p_value_one_tailed = 1 - stats.norm.cdf(z) # Display Results st.header("📈 Z-Test Results") st.write(f"**Old Design Conversion Rate:** {p1 * 100:.2f}%") st.write(f"**New Design Conversion Rate:** {p2 * 100:.2f}%") st.write(f"**Pooled Proportion:** {p:.4f}") st.write(f"**Z-Score:** {z:.3f}") st.write(f"**One-tailed p-value:** {p_value_one_tailed:.4f}") alpha = 0.05 st.write(f"**Significance Level (α):** {alpha}") # Hypothesis decision if p_value_one_tailed < alpha: st.success("🎉 **Result: Reject Null Hypothesis**") st.success("The new website design significantly improves conversions.") else: st.error("❌ **Result: Fail to Reject Null Hypothesis**") st.error("No statistically significant improvement observed.") # --- Charts Section --- st.header("📊 Visualizations") # Bar Chart fig1, ax1 = plt.subplots() versions = ["Old Design", "New Design"] conversion_rates = [p1 * 100, p2 * 100] ax1.bar(versions, conversion_rates) ax1.set_title("Comparison of Conversion Rates") ax1.set_ylabel("Conversion Rate (%)") for i, v in enumerate(conversion_rates): ax1.text(i, v + 0.2, f"{v:.2f}%", ha="center") st.pyplot(fig1) # Pie Charts fig2, (ax2, ax3) = plt.subplots(1, 2, figsize=(10, 4)) ax2.pie([x1, n1 - x1], labels=["Converted", "Not Converted"], autopct="%1.1f%%", startangle=90) ax2.set_title("Old Website Design") ax3.pie([x2, n2 - x2], labels=["Converted", "Not Converted"], autopct="%1.1f%%", startangle=90) ax3.set_title("New Website Design") st.pyplot(fig2) # Final note st.info("This Z-test helps determine if the new website design truly results in higher conversions.")