import pandas as pd

df = pd.read_csv("data/pandas/titanic.csv")
df = df.rename(columns={"Pclass": "Class", "SibSp": "Siblings"})
df[["Name", "Class", "Siblings", "Fare"]].head(3)

cheapest = df.sort_values("Fare")
cheapest[["Name", "Class", "Fare"]].head()

priciest = df.sort_values("Fare", ascending=False)
priciest[["Name", "Class", "Fare"]].head()

print(df["PassengerId"].head(3).tolist())
print(priciest["PassengerId"].head(3).tolist())

df.sort_values("fare")

print([c for c in df.columns if "fare" in c.lower()])
df.sort_values("Fare").shape

by_fare = df.copy()
result = by_fare.sort_values("Fare", inplace=True)
result.head()

by_fare = df.sort_values("Fare", ascending=False, ignore_index=True)
by_fare[["Name", "Fare"]].head(3)

by_class_age = df.sort_values(["Class", "Age"])
by_class_age[["Name", "Class", "Age"]].head()

mixed = df.sort_values(["Class", "Fare"], ascending=[True, False])
mixed[["Name", "Class", "Fare"]].head()

df.sort_values(["Class", "Fare"], ascending=[True])

fixed = df.sort_values(["Class", "Fare"], ascending=[True, False])
fixed[["Class", "Fare"]].head(3)

oldest_survivors = df.sort_values(["Survived", "Age"],
                                  ascending=[False, False])
oldest_survivors[["Name", "Survived", "Age"]].head()

by_age = df.sort_values("Age")
by_age[["Name", "Age"]].tail(3)

gaps_first = df.sort_values("Age", na_position="first")
gaps_first[["Name", "Age"]].head(3)

restored = by_age.sort_index()
restored[["PassengerId", "Name"]].head(3)

df["Embarked"].value_counts().sort_index()

df.sort_index(axis=1).columns[:6].tolist()

by_name_len = df.sort_values("Name", key=lambda s: s.str.len())
short = by_name_len["Name"].head(3)
pd.DataFrame({"name": short, "chars": short.str.len()})

longest = by_name_len["Name"].tail(1).item()
print(longest)
print(len(longest), "characters")

route = ["S", "C", "Q"]
port_type = pd.CategoricalDtype(categories=route, ordered=True)
df["Port"] = df["Embarked"].astype(port_type)
df["Port"].dtype

df.groupby("Port", observed=True)["Fare"].mean().round(2)

by_port = df.sort_values(["Port", "Fare"], ascending=[True, False])
by_port[["Name", "Port", "Fare"]].head(3)

names = {1: "Upper", 2: "Middle", 3: "Lower"}
order = {"Upper": 0, "Middle": 1, "Lower": 2}
df["Deck"] = df["Class"].map(names)
by_deck = df.sort_values("Deck", key=lambda s: s.map(order))
by_deck["Deck"].drop_duplicates().tolist()

df.nlargest(5, "Age")[["Name", "Class", "Age"]]

df.nsmallest(3, "Fare")[["Name", "Fare"]]

print((df["Fare"] == 0).sum(), "passengers paid 0")
free = df.nsmallest(3, "Fare", keep="all")
print(len(free), "rows with keep=all")

third = df[df["Class"] == 3]
third.nlargest(3, "Fare")[["Name", "Siblings", "Fare"]]

ages = df["Age"].head(5)
pd.DataFrame({"age": ages,
              "average": ages.rank(ascending=False),
              "min": ages.rank(ascending=False, method="min"),
              "dense": ages.rank(ascending=False, method="dense"),
              "first": ages.rank(ascending=False, method="first")})

df["Fare_rank"] = df["Fare"].rank(ascending=False, method="min")
cols = ["Name", "Fare", "Fare_rank"]
df.sort_values("Fare_rank")[cols].head()

dense = df["Fare"].rank(ascending=False, method="dense")
print(dense.max(), "dense rank levels")
print(df["Fare"].nunique(), "different fares")

df["Fare_pct"] = df["Fare"].rank(pct=True).round(2)
df[["Name", "Fare", "Fare_pct"]].head()

grp = df.groupby("Class")["Fare"]
df["Class_rank"] = grp.rank(ascending=False, method="min")
top1 = df[df["Class_rank"] == 1]
print(len(top1), "passengers share a rank of 1")
top1.drop_duplicates("Class")[["Name", "Class", "Fare"]]

age_rank = df.groupby("Sex")["Age"].rank(ascending=False,
                                         method="min")
women = df[(df["Sex"] == "female") & (age_rank == 1)]
women[["Name", "Class", "Age"]]

top1.drop_duplicates("Class")[["Class", "Fare", "Fare_pct"]]

gap = pd.read_csv("data/pandas/gapminder.csv")
wide = gap.pivot(index="country", columns="year", values="lifeExp")
life = wide[[1952, 2007]]
life.sort_values(2007, ascending=False).head(3).round(1)

ranks = life.rank(ascending=False, method="min").astype(int)
ranks["climb"] = ranks[1952] - ranks[2007]
ranks.sort_values(2007).head(3)

ranks.nlargest(5, "climb")

ranks.nsmallest(5, "climb")

top = ranks.nlargest(10, "climb")
cont = gap.drop_duplicates("country").set_index("country")
cont.loc[top.index, "continent"].value_counts()

import matplotlib.pyplot as plt

fig, ax = plt.subplots(figsize=(8, 4))
top["climb"].sort_values().plot.barh(ax=ax, color="seagreen")
ax.set_title("Biggest climbs in life expectancy rank, 1952-2007")
ax.set_xlabel("Places gained")
ax.set_ylabel("Country")
plt.show()

latest = gap[gap["year"] == 2007].copy()
by_cont = latest.groupby("continent")["lifeExp"]
latest["cont_rank"] = by_cont.rank(ascending=False, method="min")
leaders = latest[latest["cont_rank"] == 1]
leaders[["continent", "country", "lifeExp"]].sort_values("lifeExp")