import pandas as pd
df = pd.read_csv("data/pandas/tips.csv")
print(df.shape)
df.head()

(df
    .query("time == 'Lunch'")
    .sort_values("tip", ascending=False)
    .head(3))

df.info()

df.describe().round(2)

df.value_counts(["day", "time"])

print(df.size)
print(df["size"].head(3).tolist())

print(df.duplicated().sum())
df[df.duplicated(keep=False)]

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

df["per_person"] = (df["total_bill"] / df["size"]).round(2)
df[["total_bill", "size", "per_person"]].head(3)

df["tip_pct"].describe().round(1)

if df["tip_pct"] > 20:
    print("generous")

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

df["tip_each"] = (df["tip"] / df["size"]).round(2)
df.nlargest(3, "tip_each")[["tip", "size", "tip_each"]]

print(df["day"].unique().tolist())
names = {"Thur": "Thursday", "Fri": "Friday",
         "Sat": "Saturday", "Sun": "Sunday"}
df["day_name"] = df["day"].map(names)
print(df["day_name"].value_counts())

only_wknd = {"Sat": "Saturday", "Sun": "Sunday"}
print(df["day"].map(only_wknd).isna().sum())

labels = {"Yes": "Smoker", "No": "Non-smoker"}
df["smoker"] = df["smoker"].replace(labels)
df["smoker"].value_counts()

bands = [0, 10, 15, 20, 100]
band_names = ["Low", "Fair", "Good", "Generous"]
df["tip_band"] = pd.cut(df["tip_pct"], bins=bands,
                        labels=band_names)
df["tip_band"].value_counts(sort=False)

def party_type(n):
    return "Group" if n >= 4 else "Small"

df["party"] = df["size"].apply(party_type)
df["party"].value_counts()

df["size"].apply(party_type())

df["size"].apply(party_type).head(3)

import numpy as np
big = df["total_bill"] > 30
df["bill_size"] = np.where(big, "Big", "Regular")
df["bill_size"].value_counts()

cols = ["total_bill", "tip", "tip_pct", "day", "time"]
generous = df[df["tip_pct"] > 20]
print(len(generous))
generous[cols].head()

df[df["tip_pct"] > 20 and df["time"] == "Dinner"]

mask = (df["tip_pct"] > 20) & (df["time"] == "Dinner")
print(mask.sum())

df.query("tip_pct > 20 and time == 'Dinner'").shape

weekend = df[df["day"].isin(["Sat", "Sun"])]
weekday = df[~df["day"].isin(["Sat", "Sun"])]
print(len(weekend), len(weekday))

print(round(weekend["tip_pct"].mean(), 2))
print(round(weekday["tip_pct"].mean(), 2))

sat = df.query("day == 'Sat' and size >= 3")
low = sat.query("tip_pct < 10")
print(len(low), len(sat))

raw = pd.read_csv("data/pandas/tips.csv")
print(raw.columns.tolist())

raw.assign(
    tip_pct=raw["tip"] / raw["total_bill"] * 100,
    high=raw["tip_pct"] > 20,
)

tidy = raw.assign(
    tip_pct=lambda d: d["tip"] / d["total_bill"] * 100,
    high=lambda d: d["tip_pct"] > 20,
)
tidy[["tip", "tip_pct", "high"]].head(3).round(1)

report = (
    raw.assign(tip_pct=lambda d: d["tip"] / d["total_bill"] * 100)
    .query("size >= 4")
    .sort_values("tip_pct", ascending=False)
    .loc[:, ["day", "size", "total_bill", "tip_pct"]]
    .round(1)
)
report.head()

def add_tip_cols(d):
    d = d.assign(tip_pct=d["tip"] / d["total_bill"] * 100)
    return d.assign(tip_each=d["tip"] / d["size"])

raw.pipe(add_tip_cols).nlargest(3, "tip_each").round(2)

top = (
    raw.query("time == 'Dinner'")
    .assign(per_person=lambda d: d["total_bill"] / d["size"])
    .nlargest(3, "per_person")
)
top[["day", "size", "total_bill", "per_person"]].round(2)

is_smoker = df["smoker"] == "Smoker"
smoke = df.loc[is_smoker, "tip_pct"]
other = df.loc[~is_smoker, "tip_pct"]
print(round(smoke.mean(), 2), round(other.mean(), 2))

print(smoke.median(), other.median())

show = ["total_bill", "tip", "tip_pct", "smoker"]
df.nlargest(3, "tip_pct")[show]

print(round(smoke[smoke < 50].mean(), 2))
print(round(smoke.std(), 2), round(other.std(), 2))

spend = df.groupby("smoker")["per_person"].median()
spend

import matplotlib.pyplot as plt
ax = df.boxplot(column="tip_pct", by="smoker", figsize=(7, 4))
plt.suptitle("")
ax.set_title("Tip percentage: smokers vs non-smokers")
ax.set_xlabel("Table")
ax.set_ylabel("Tip as % of bill")
plt.show()

gap = smoke.median() - other.median()
print(f"Median gap: {gap:.1f} points")