# my_app.py

import streamlit as st
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

# Load dataset
df = sns.load_dataset("tips")
df["tip_pct"] = (df["tip"] / df["total_bill"]) * 100
df["tip_pct"] = df["tip_pct"].round(2)

# App title
st.title("Restaurant Tip Insights")
st.markdown("Explore tip percentages by day, gender, and smoker status.")

# Sidebar filters
day = st.sidebar.selectbox("Select a day", df['day'].unique())
gender = st.sidebar.multiselect("Select gender(s)", df['sex'].unique(), default=df['sex'].unique())
smoker = st.sidebar.multiselect("Select smoker status", df['smoker'].unique(), default=df['smoker'].unique())

# Filter data
filtered = df[(df['day'] == day) & (df['sex'].isin(gender)) & (df['smoker'].isin(smoker))]

# KPI
avg_tip = filtered['tip_pct'].mean()
st.metric(label=f"Average Tip % on {day}", value=f"{avg_tip:.2f}%")

median_smokers = filtered[filtered["smoker"]=="Yes"]["tip_pct"].median()
median_non_smokers = filtered[filtered["smoker"]=="No"]["tip_pct"].median()
st.write(f"Median tip % for smokers: {median_smokers:.2f}%")
st.write(f"Median tip % for non-smokers: {median_non_smokers:.2f}%")

# Plot 1 — Average Tip Percentage by Day
avg_by_day = df.groupby("day")["tip_pct"].mean()
fig1, ax1 = plt.subplots(figsize=(6,4))
avg_by_day.plot(kind="bar", color="skyblue", ax=ax1)
ax1.set_title("Average Tip Percentage by Day")
ax1.set_xlabel("Day of Week")
ax1.set_ylabel("Average Tip Percentage (%)")
ax1.grid(axis="y", linestyle="--", alpha=0.7)
st.pyplot(fig1)
st.markdown("**Interpretation:** Friday has the highest average tip percentage, Saturday the lowest.")

# Plot 2 — Tip Percentage vs Total Bill by Sex
fig2, ax2 = plt.subplots(figsize=(8,6))
sns.scatterplot(x="total_bill", y="tip_pct", hue="sex", data=filtered, alpha=0.6, ax=ax2)
ax2.set_title("Tip Percentage vs Total Bill by Sex")
ax2.set_xlabel("Total Bill")
ax2.set_ylabel("Tip Percentage (%)")
ax2.grid(True, linestyle="--", alpha=0.5)
st.pyplot(fig2)
st.markdown("**Interpretation:** Tip % tends to decrease as total bill increases. Some small bills have very high tip percentages.")

# Plot 3 — Tip Percentage by Smoking Status
fig3, ax3 = plt.subplots()
sns.boxplot(x="smoker", y="tip_pct", data=filtered, palette="Set2", ax=ax3)
ax3.set_title("Tip Percentage by Smoking Status")
ax3.set_xlabel("Smoker Status")
ax3.set_ylabel("Tip Percentage (%)")
ax3.grid(axis="y", linestyle="--", alpha=0.7)
st.pyplot(fig3)
st.markdown("**Interpretation:** Smoking status does not largely impact tip %, but some smokers tip higher.")