import pandas as pd
tips = pd.read_csv("data/pandas/tips.csv")
print(tips.shape, tips.size)
print(tips.columns.tolist())

tips["bill_after_tip"] = tips["total_bill"] + tips["tip"]
tips[["total_bill", "tip", "bill_after_tip"]].head(3)

first = tips.head(4)
log = pd.DataFrame({
    "day": first["day"].to_list(),
    "size": first["size"].to_list(),
    "tip": first["tip"].to_list(),
})
log

log["per_person"] = (log["tip"] / log["size"]).round(2)
log

log["group"] = log["size"] >= 3
log[["size", "tip", "group"]]

extra = tips.loc[[4], ["day", "size"]]
log = pd.concat([log, extra], ignore_index=True)
log

log["Size"].sum()

print(log.columns.tolist())
print(log["size"].sum())

tips = tips.drop(columns="bill_after_tip")
tips.head()

tips.info()

print(tips.isna().sum().sum())

tips.nunique()

tips.sample(5, random_state=42)

tips["day"].value_counts()

tips["time"].value_counts()

tips.describe().round(2)

tips.describe(exclude="number")

tips["size"].value_counts().sort_index()

tips["smoker"].value_counts(normalize=True).round(3)

print(tips["tip"].median())
print(tips["tip"].max())

import matplotlib.pyplot as plt
ax = tips["tip"].plot(kind="hist", bins=20, figsize=(7, 4))
ax.set_title("Tips left on 244 restaurant bills")
ax.set_xlabel("Tip (dollars)")
ax.set_ylabel("Number of bills")
plt.show()

tips["tip_pct"] = tips["tip"] / tips["total_bill"] * 100
tips[["total_bill", "tip", "tip_pct"]].head().round(1)

tips["per_person"] = tips["total_bill"] / tips["size"]
print(round(tips["per_person"].median(), 2))
print(round(tips["tip_pct"].median(), 2))

tips["big_party"] = tips["size"] >= 3
tips["big_party"].value_counts()

pp_cols = ["total_bill", "size", "per_person"]
tips.nlargest(3, "per_person")[pp_cols].round(2)

generous = tips["tip_pct"] >= 20
print(generous.sum(), round(generous.mean(), 3))

cols = ["total_bill", "tip", "size", "day", "tip_pct"]
tips.nlargest(3, "tip_pct")[cols].round(1)

dinner = tips[tips["time"] == "Dinner"]
print(len(dinner), round(dinner["tip_pct"].median(), 2))

tips[(tips["day"] == "Sat") and (tips["smoker"] == "Yes")]

sat_smoke = (tips["day"] == "Sat") & (tips["smoker"] == "Yes")
print(sat_smoke.sum())

fri_sun = (tips["day"] == "Fri") | (tips["day"] == "Sun")
print(fri_sun.sum())

by_bill = tips.sort_values("total_bill", ascending=False)
by_bill[cols].head(3).round(1)

two_keys = tips.sort_values(["day", "tip_pct"],
                            ascending=[True, False])
two_keys[cols].head(3).round(1)

lunch = tips[tips["time"] == "Lunch"]
lunch.nlargest(3, "tip_pct")[cols].round(1)

ax = tips.plot(kind="scatter", x="total_bill", y="tip",
                figsize=(7, 4), alpha=0.6)
ax.set_title("Tip against bill for 244 tables")
ax.set_xlabel("Total bill (dollars)")
ax.set_ylabel("Tip (dollars)")
plt.show()

tips["total_bill"].corr("tip")

print(round(tips["total_bill"].corr(tips["tip"]), 3))
print(round(tips["size"].corr(tips["tip_pct"]), 3))

counts = tips["day"].value_counts()
ax = counts.plot(kind="bar", rot=0, figsize=(7, 4))
ax.set_title("Bills per day")
ax.set_xlabel("Day")
ax.set_ylabel("Number of bills")
plt.show()

big = tips["big_party"]
print(round(tips.loc[big, "tip"].median(), 2),
      round(tips.loc[~big, "tip"].median(), 2))

big_mean = tips.loc[big, "tip_pct"].mean()
small_mean = tips.loc[~big, "tip_pct"].mean()
print(round(big_mean, 2), round(small_mean, 2))

by_size = tips.groupby("size")["tip_pct"].agg(["count", "median"])
by_size.round(1)

big_med = tips.loc[big, "tip_pct"].median()
small_med = tips.loc[~big, "tip_pct"].median()
print(round(big_med, 2), round(small_med, 2))

normal = tips["tip_pct"] < 40
t_big = tips.loc[big & normal, "tip_pct"].mean()
t_small = tips.loc[~big & normal, "tip_pct"].mean()
print((~normal).sum(), round(t_big, 2), round(t_small, 2))

summary = pd.DataFrame({"mean": [small_mean, big_mean],
                        "median": [small_med, big_med]},
                       index=["1-2 people", "3+ people"])
ax = summary.plot(kind="bar", rot=0, figsize=(7, 4))
ax.set_title("Tip percentage by party size")
ax.set_xlabel("Party")
ax.set_ylabel("Tip (% of bill)")
plt.show()

gap_mean = small_mean - big_mean
gap_med = small_med - big_med
print(f"Gap: {gap_mean:.1f} pts by mean, {gap_med:.1f} by median")