Series

import pandas as pd
import numpy as np

# ===== Creation =====
s = pd.Series([10, 20, 30, 40], index=['a', 'b', 'c', 'd'])
print(s)

# Default integer index:
s2 = pd.Series([1, 2, 3])
# 0    1
# 1    2
# 2    3

# From a dict (key -> index, value -> values):
s3 = pd.Series({'a': 1, 'b': 2, 'c': 3})

# ===== Attributes =====
s.dtype       # int64
s.index       # Index(['a', 'b', 'c', 'd'])
s.values      # np.array([10, 20, 30, 40])
s.shape       # (4,)
s.name = 'amount'

# ===== Indexing =====
s['a']         # 10  (label)
s.iloc[0]      # 10  (position)
s[['a', 'c']]  # subset

# Slicing — labels INCLUSIVE on both ends:
s['a':'c']

# Boolean mask:
s[s > 15]

# ===== Vectorised math =====
s + 1
s * 2
s ** 0.5
np.log(s)
s + pd.Series([100, 200], index=['a', 'b'])
# Adds where index matches; NaN elsewhere.

# ===== Stats =====
s.mean(), s.std(), s.min(), s.max()
s.sum(), s.cumsum(), s.cummax()
s.quantile(0.95)
s.describe()

# ===== Missing data =====
s4 = pd.Series([1.0, np.nan, 3.0])
s4.isna()        # [F, T, F]
s4.fillna(0)
s4.dropna()
s4.interpolate()

# ===== String operations (.str accessor) =====
names = pd.Series(['Alex', 'Sam', 'Lee'])
names.str.lower()
names.str.contains('a', case=False)
names.str.len()

# ===== Date operations (.dt accessor) =====
dates = pd.to_datetime(pd.Series(['2024-04-10', '2024-04-11']))
dates.dt.year
dates.dt.day_name()
dates.dt.tz_localize('Australia/Sydney')

# ===== Categorical for low-cardinality strings =====
cats = pd.Series(['a', 'b', 'a', 'c'], dtype='category')
cats.cat.categories
cats.memory_usage()

# ===== map vs apply =====
# map: per-element transform, simple
s.map(lambda x: x * 10)
s.map({'a': 'A', 'b': 'B'})       # also lookup-table style

# apply: more flexible, also per-element
s.apply(lambda x: x ** 2)

# ===== Conversion to + from numpy =====
arr = s.to_numpy()
s5 = pd.Series(arr, index=['a','b','c','d'], name='amt')

# ===== Patterns to internalise =====
# - dtype matters: int64 / float64 / category / datetime64[ns]
# - .iloc for position, [] / .loc for labels
# - Use .str / .dt accessors over Python loops
# - Convert low-cardinality strings to 'category' to save memory

# ===== Pitfalls =====
# - Mixing label and position indexing
# - Implicit alignment when adding series with different indexes -> surprise NaNs
# - SettingWithCopyWarning when assigning into a sliced view
# - .apply with a Python function on a hot column -> slow; reach for vectorised

import pandas as pd
s = pd.Series([10, 20, 30], index=['a','b','c'])
print(s)
print(s['b'])              # 20