Advanced Pandas — Complete Notes¶

CampusX Session 19–22 (GroupBy + Merging/Joining/Concatenating + MultiIndex + Vectorized String Ops & DateTime + Pivot Table)¶

SESSION 19 — GROUPBY OBJECT IN PANDAS¶

1. GroupBy hai kya?¶

GroupBy ka concept Split → Apply → Combine hai:

Split — data ko groups mein baant do (kisi column ke basis pe)
Apply — har group pe koi function lagao (sum, mean, count...)
Combine — results ko wapas ek result mein jod do

SQL ka GROUP BY yaad hai? Bilkul wahi cheez hai.

import pandas as pd
import numpy as np

movies = pd.read_csv('imdb-top-1000.csv')

2. GroupBy object banana¶

genres = movies.groupby('Genre')
print(genres)
# <pandas.core.groupby.generic.DataFrameGroupBy object at 0x...>

Note: Ye khud kuch print nahi karta — ye ek "lazy" object hai. Aggregation lagane pe result deta hai.

3. Built-in Aggregation Functions¶

genres.sum(numeric_only=True)
genres.mean(numeric_only=True)
genres.min()
genres.max()
genres.count()
genres.std()
genres.var()
genres.median()
genres.first()      # har group ki pehli row
genres.last()       # har group ki aakhri row
genres.nth(5)       # har group ki 6th row
genres.size()       # har group mein kitni rows ⭐
genres.describe()
genres.sample()
genres.nunique()

Practical examples:

# Har genre ki total gross earning
movies.groupby('Genre')['Gross'].sum().sort_values(ascending=False)

# Har genre mein kitni movies
movies.groupby('Genre').size().sort_values(ascending=False)

# Har genre ki average IMDB rating
movies.groupby('Genre')['IMDB_Rating'].mean().sort_values(ascending=False)

# Top 3 genres by number of movies
movies['Genre'].value_counts().head(3)

4. GroupBy Attributes & Methods¶

4.1 `len()` — kitne groups bane¶

len(movies.groupby('Genre'))       # kitne unique genres
movies['Genre'].nunique()          # same answer

4.2 `size()` — har group ka size¶

movies.groupby('Genre').size()

4.3 `first()`, `last()`, `nth()`¶

genres.first()
genres.last()
genres.nth(6)

4.4 `get_group()` ⭐¶

Ek particular group ka poora data nikaalo.

genres.get_group('Fantasy')
# Barabar hai:
movies[movies['Genre'] == 'Fantasy']

4.5 `groups` attribute¶

genres.groups
# Dictionary: {group_name: [index positions]}

4.6 `describe()`¶

genres.describe()

4.7 `sample()`¶

genres.sample(2, replace=True)      # har group se 2 random rows

4.8 `nunique()`¶

genres.nunique()      # har group mein har column ki unique values

5. `agg()` — Multiple aggregations ek saath ⭐⭐¶

5.1 List pass karo¶

genres.agg(['min', 'max', 'mean', 'sum'])

5.2 Dictionary pass karo (alag column, alag function)¶

genres.agg({
    'Runtime': 'mean',
    'IMDB_Rating': 'mean',
    'No_of_Votes': 'sum',
    'Gross': 'sum',
    'Metascore': 'min'
})

5.3 Named aggregation (naye syntax)¶

movies.groupby('Genre').agg(
    avg_rating=('IMDB_Rating', 'mean'),
    total_gross=('Gross', 'sum'),
    movie_count=('Series_Title', 'count')
)

6. GroupBy pe loop chalana¶

for group_name, group_df in movies.groupby('Genre'):
    print(group_name)
    print(group_df.shape)
    print('---')

Practical example — har genre ki top-rated movie:

df = pd.DataFrame(columns=movies.columns)

for group, data in movies.groupby('Genre'):
    df = pd.concat([df, data.sort_values('IMDB_Rating', ascending=False).head(1)])

7. `apply()` on GroupBy ⭐¶

Custom function har group pe apply karta hai.

# Har genre ke andar har column ki unique value count
def foo(group):
    return group['IMDB_Rating'].max() - group['IMDB_Rating'].min()

movies.groupby('Genre').apply(foo)

Popular example — har group ke andar percentage nikalna:

def rating_percent(group):
    group['rating_share'] = group['IMDB_Rating'] / group['IMDB_Rating'].sum()
    return group

movies.groupby('Genre').apply(rating_percent)

8. Multiple columns pe GroupBy ⭐¶

duo = movies.groupby(['Director', 'Star1'])
duo.size()
duo['Gross'].sum().sort_values(ascending=False).head(5)

# Har director-actor jodi ki sabse zyada earning wali movie
duo['Gross'].max()

Result ek MultiIndex Series hota hai (Session 21 mein detail).

# Access karna
duo.get_group(('Aamir Khan', 'Amole Gupte'))

9. GroupBy — Practical Questions (Colab exercises)¶

# 1. Sabse zyada rated genre
movies.groupby('Genre')['IMDB_Rating'].mean().sort_values(ascending=False).head(1)

# 2. Har genre ka top director (by number of movies)
movies.groupby(['Genre', 'Director']).size().sort_values(ascending=False)

# 3. Har actor ki highest rated movie
movies.groupby('Star1')['IMDB_Rating'].max().sort_values(ascending=False).head(5)

# 4. Har genre mein kitne unique directors
movies.groupby('Genre')['Director'].nunique().sort_values(ascending=False)

# 5. Har genre ka runtime average
movies.groupby('Genre')['Runtime'].mean().sort_values(ascending=False)

---¶

SESSION 20 — MERGING, JOINING & CONCATENATING¶

1. `pd.concat()` ⭐¶

Do ya zyada DataFrames ko jodna (stack karna).

1.1 Vertically (axis=0, default) — rows add¶

courses = pd.read_csv('courses.csv')
students = pd.read_csv('students.csv')

pd.concat([df1, df2])                     # ek ke neeche ek
pd.concat([df1, df2], ignore_index=True)  # index reset kar do ⭐

1.2 Horizontally (axis=1) — columns add¶

pd.concat([df1, df2], axis=1)

1.3 `keys` — MultiIndex banana¶

multi = pd.concat([df1, df2], keys=['dec', 'nov'])
multi.loc['dec']            # sirf december ka data
multi.loc[('nov', 3)]       # november ka index 3

1.4 Sirf common columns rakhna¶

pd.concat([df1, df2], join='inner')     # sirf common columns
pd.concat([df1, df2], join='outer')     # sab columns (default), missing → NaN

2. `pd.merge()` ⭐⭐⭐¶

SQL ke JOIN jaisa. Common column ke basis pe do tables jodna.

Syntax¶

pd.merge(left_df, right_df, how='inner', on='common_column')

2.1 Types of joins¶

`how`	Kya karta hai
`'inner'`	Sirf wo rows jo dono mein hain (default)
`'left'`	Left ki saari rows + matching right
`'right'`	Right ki saari rows + matching left
`'outer'`	Dono ki saari rows (union)
`'cross'`	Cartesian product

students.merge(regs, how='inner', on='student_id')
students.merge(regs, how='left', on='student_id')
students.merge(regs, how='right', on='student_id')
students.merge(regs, how='outer', on='student_id')

2.2 Alag naam wale columns pe merge¶

pd.merge(df1, df2, left_on='id', right_on='student_id')

2.3 Index pe merge¶

pd.merge(df1, df2, left_index=True, right_index=True)

2.4 Suffixes (jab column names clash karein)¶

pd.merge(df1, df2, on='id', suffixes=('_left', '_right'))

2.5 `indicator`¶

pd.merge(df1, df2, how='outer', on='id', indicator=True)
# Ek '_merge' column aata hai: 'both', 'left_only', 'right_only'

3. Relationship Types¶

Type	Example
One to One	Har student ka ek hi roll number
One to Many	Ek course mein kai students
Many to Many	Kai students, kai courses (registration table se)

4. `df.join()`¶

Index pe merge karne ka shortcut.

df1.join(df2)                       # index pe left join
df1.join(df2, how='inner')
df1.join(df2, on='key_column')

5. Merging — Practical Questions (Colab exercises)¶

Datasets: courses.csv, students.csv, nov.csv, dec.csv, matches.csv, deliveries.csv

# 1. Kitne students ne kabhi enroll nahi kiya
students.merge(regs, how='left', on='student_id')['name'].isnull().sum()

# 2. Kaunse course mein koi student nahi
courses.merge(regs, how='left', on='course_id')['student_id'].isnull().sum()

# 3. Total revenue generated
regs.merge(courses, how='inner', on='course_id')['price'].sum()

# 4. Month by month revenue
temp = pd.concat([nov, dec], keys=['Nov', 'Dec']).reset_index()
temp.merge(courses, on='course_id').groupby('level_0')['price'].sum()

# 5. Har course ka revenue
regs.merge(courses, on='course_id').groupby('course_name')['price'].sum().sort_values(ascending=False)

# 6. Sabse zyada revenue dene wala student
regs.merge(students, on='student_id').merge(courses, on='course_id') \
    .groupby('name')['price'].sum().sort_values(ascending=False).head(1)

# 7. Har course ka average rating
# 8. Kaunsa course sabse popular hai
regs.merge(courses, on='course_id')['course_name'].value_counts()

# 9. IPL: har batsman ka total run
deliveries.merge(matches, left_on='match_id', right_on='id') \
    .groupby('batter')['batsman_runs'].sum().sort_values(ascending=False)

# 10. Kohli ne kis stadium mein sabse zyada run banaye
new = deliveries.merge(matches, left_on='match_id', right_on='id')
new[new['batter'] == 'V Kohli'].groupby('venue')['batsman_runs'].sum() \
   .sort_values(ascending=False).head(1)

---¶

SESSION 21 — MULTIINDEX SERIES & DATAFRAMES¶

1. MultiIndex kyun chahiye?¶

Kabhi kabhi humein higher-dimensional data ko 2D table mein represent karna hota hai. MultiIndex (Hierarchical Index) se ek se zyada level ka index bana sakte hain.

Example: Har company ke har saal ka data → index = (year, company)

2. MultiIndex banana¶

2.1 `from_tuples()`¶

index_val = [('cse', 2019), ('cse', 2020), ('cse', 2021), ('cse', 2022),
             ('ece', 2019), ('ece', 2020), ('ece', 2021), ('ece', 2022)]

multiindex = pd.MultiIndex.from_tuples(index_val)
multiindex.levels[0]     # ['cse', 'ece']
multiindex.levels[1]     # [2019, 2020, 2021, 2022]

2.2 `from_product()` (aasan tareeqa)¶

pd.MultiIndex.from_product([['cse', 'ece'], [2019, 2020, 2021, 2022]])

3. MultiIndex Series¶

s = pd.Series([1, 2, 3, 4, 5, 6, 7, 8], index=multiindex)
s['cse']              # cse ka poora data
s['cse'][2019]        # ek value
s[('cse', 2019)]      # same

4. `unstack()` aur `stack()` ⭐¶

unstack() → inner index level ko columns bana deta hai (Series → DataFrame)
stack() → columns ko wapas index bana deta hai (DataFrame → Series)

temp = s.unstack()      # 2D table ban gaya
temp.stack()            # wapas MultiIndex Series

Ye ek dusre ka inverse hain.

5. MultiIndex DataFrame¶

branch_df = pd.DataFrame(
    [
        [1, 2], [3, 4], [5, 6], [7, 8],
        [9, 10], [11, 12], [13, 14], [15, 16]
    ],
    index=multiindex,
    columns=['avg_package', 'students']
)

Columns pe bhi MultiIndex¶

branch_df = pd.DataFrame(
    [[1,2,0,0],[3,4,0,0],[5,6,0,0],[7,8,0,0],
     [9,10,0,0],[11,12,0,0],[13,14,0,0],[15,16,0,0]],
    index=multiindex,
    columns=pd.MultiIndex.from_product([['delhi','mumbai'], ['avg_package','students']])
)

Ab rows aur columns dono MultiIndex hain.

6. MultiIndex DataFrame ke saath kaam¶

branch_df.head()
branch_df.shape
branch_df.info()
branch_df.duplicated().sum()
branch_df.isnull().sum()

# Extracting rows
branch_df.loc['cse']
branch_df.loc[('cse', 2022)]
branch_df.loc[('cse', 2019):('ece', 2020)]

# Extracting columns
branch_df['delhi']
branch_df['delhi']['students']
branch_df.iloc[:, 1:3]

# Transpose
branch_df.transpose()

`sort_index()` on MultiIndex ⭐¶

branch_df.sort_index(ascending=False)
branch_df.sort_index(ascending=[False, True])
branch_df.sort_index(level=0, ascending=False)

`swaplevel()`¶

branch_df.swaplevel(axis=0)      # index levels swap
branch_df.swaplevel(axis=1)      # column levels swap

7. Long vs Wide Data ⭐⭐¶

Wide Format	Long Format
Har variable ka apna column	Ek 'variable' aur ek 'value' column
Insaan ke liye padhne mein aasan	Machines/plotting ke liye better
Excel jaisa	Database jaisa

Example — Wide:

branch   2019   2020   2021
cse       1      3      5
ece       9     11     13

Example — Long:

branch   year   students
cse      2019      1
cse      2020      3
cse      2021      5
ece      2019      9

8. `melt()` — Wide se Long ⭐¶

pd.melt(df, id_vars=['branch'], var_name='year', value_name='students')

Real example — COVID data:

death = pd.read_csv('time_series_covid19_deaths_global.csv')
confirm = pd.read_csv('time_series_covid19_confirmed_global.csv')

death = death.melt(id_vars=['Province/State','Country/Region','Lat','Long'],
                   var_name='date', value_name='num_deaths')

confirm = confirm.melt(id_vars=['Province/State','Country/Region','Lat','Long'],
                       var_name='date', value_name='num_cases')

final = confirm.merge(death, on=['Province/State','Country/Region','Lat','Long','date'])

# Ab analysis aasan:
final.groupby('Country/Region')['num_cases'].max().sort_values(ascending=False)

9. `pivot_table()` ⭐⭐⭐¶

Long data ko Wide mein badalne ka aur summary banane ka sabse powerful tool. Excel ke Pivot Table jaisa.

Syntax¶

df.pivot_table(index=..., columns=..., values=..., aggfunc=...)

import seaborn as sns
df = sns.load_dataset('tips')

# Har gender ka har day pe average bill
df.pivot_table(index='sex', columns='day', values='total_bill')

# aggfunc badlo
df.pivot_table(index='sex', columns='day', values='total_bill', aggfunc='sum')
df.pivot_table(index='sex', columns='day', values='total_bill', aggfunc='count')
df.pivot_table(index='sex', columns='day', values='total_bill', aggfunc='std')

# Multiple index / columns
df.pivot_table(index=['sex', 'smoker'], columns=['day', 'time'],
               values='total_bill', aggfunc='mean')

# margins (Grand Total row/column)
df.pivot_table(index='sex', columns='day', values='total_bill',
               aggfunc='sum', margins=True)

# Multiple values
df.pivot_table(index='sex', columns='day', values=['total_bill', 'tip'], aggfunc='mean')

# Missing values
df.pivot_table(index='sex', columns='day', values='tip', fill_value=0)

Plotting a pivot table ⭐¶

expense = pd.read_csv('expense_data.csv')
expense['Date'] = pd.to_datetime(expense['Date'])
expense['month'] = expense['Date'].dt.month_name()

expense.pivot_table(index='month', columns='Category',
                    values='INR', aggfunc='sum', fill_value=0).plot()

pivot() vs pivot_table(): - pivot() — koi aggregation nahi karta, duplicate index-column combination pe error deta hai - pivot_table() — aggregation karta hai, duplicates handle kar leta hai ✅

---¶

SESSION 22 — VECTORIZED STRING OPERATIONS & DATETIME¶

PART A — VECTORIZED STRING OPERATIONS (`.str`)¶

1. Problem kya hai?¶

Normal Python mein string functions ek string pe chalte hain. Poore column pe chalane ke liye loop lagana padta — slow.

Pandas mein .str accessor hai jo poore column pe vectorized string operation karta hai.

s = pd.Series(['cat', 'mat', None, 'rat'])
s.str.startswith('c')

Setup:

df = pd.read_csv('titanic.csv')

2. Common `.str` methods¶

2.1 Case change¶

df['Name'].str.lower()
df['Name'].str.upper()
df['Name'].str.capitalize()      # Pehla letter capital
df['Name'].str.title()           # Har word ka pehla letter capital
df['Name'].str.swapcase()

2.2 `len()`¶

df['Name'].str.len()                          # naam kitna lamba
df[df['Name'].str.len() == df['Name'].str.len().max()]   # sabse lamba naam

2.3 `strip()` — whitespace hatana¶

df['Name'].str.strip()      # dono taraf se
df['Name'].str.lstrip()     # left
df['Name'].str.rstrip()     # right

2.4 `split()` aur `get()` ⭐¶

# Titanic naam format: "Braund, Mr. Owen Harris"
df['lastname'] = df['Name'].str.split(',').str.get(0)
df['title'] = df['Name'].str.split(',').str.get(1).str.strip().str.split('.').str.get(0)

df['title'].value_counts()
# Mr, Miss, Mrs, Master, Dr, Rev...

# expand=True se alag columns mil jaate hain
df['Name'].str.split(',', expand=True)

2.5 `replace()` ⭐¶

df['title'] = df['title'].str.replace('Ms', 'Miss')
df['title'] = df['title'].str.replace('Mlle', 'Miss')

2.6 Filtering ke liye string methods ⭐¶

df[df['Name'].str.startswith('A')]
df[df['Name'].str.endswith('n')]
df[df['Name'].str.contains('Mr')]
df[df['Name'].str.contains('John|James')]     # regex OR

2.7 Check functions¶

df['Name'].str.isdigit()
df['Name'].str.isalpha()
df['Name'].str.isalnum()
df['Name'].str.isupper()
df['Name'].str.islower()
df['Name'].str.isspace()

2.8 Slicing¶

df['Name'].str[::-1]        # reverse
df['Name'].str[0:5]         # pehle 5 characters
df['Name'].str[0]           # pehla character

2.9 `cat()` — join karna¶

df['Name'].str.cat(sep=', ')

2.10 Padding¶

df['Name'].str.pad(width=50, side='left', fillchar='*')
df['Name'].str.zfill(10)

3. Regex ke saath `.str` ⭐¶

# Jinke naam mein 'chidambaram' pattern hai
df[df['Name'].str.contains('^[A-Z][a-z]+$', regex=True)]

# Extract karna
df['Name'].str.extract(r'([A-Za-z]+)\.')      # title extract

# Kitni baar pattern aaya
df['Name'].str.count('a')

# Find
df['Name'].str.find('Mr')

# findall
df['Name'].str.findall(r'[A-Z]')

PART B — DATETIME IN PANDAS¶

1. `Timestamp` object¶

Pandas ka datetime object.

# Alag alag tareeqe
pd.Timestamp('2023/1/5')
pd.Timestamp('2023-1-5')
pd.Timestamp('2023, 1, 5')
pd.Timestamp('5th January 2023')
pd.Timestamp('5th January 2023 9:21AM')

import datetime as dt
pd.Timestamp(dt.datetime(2023, 1, 5, 9, 21, 56))

Attributes nikalna¶

x = pd.Timestamp('2023/1/5')
x.year        # 2023
x.month       # 1
x.day         # 5
x.hour
x.minute
x.second
x.dayofweek   # 0 = Monday
x.day_name()  # 'Thursday'
x.month_name()
x.is_month_start
x.is_month_end
x.quarter

Current time¶

pd.Timestamp('today')
pd.Timestamp('now')
dt.datetime.now()

2. `DatetimeIndex`¶

dt_index = pd.DatetimeIndex(['2023/1/1', '2022/1/1', '2021/1/1'])
pd.Series([1, 2, 3], index=dt_index)

3. `date_range()` ⭐¶

Dates ki series generate karta hai.

pd.date_range(start='2023/1/5', end='2023/2/28', freq='D')     # daily
pd.date_range(start='2023/1/5', end='2023/2/28', freq='3D')    # har 3rd din
pd.date_range(start='2023/1/5', end='2023/2/28', freq='B')     # business days
pd.date_range(start='2023/1/5', end='2023/2/28', freq='W')     # weekly (Sunday)
pd.date_range(start='2023/1/5', end='2023/2/28', freq='W-THU') # har Thursday
pd.date_range(start='2023/1/5', end='2023/2/28', freq='h')     # hourly
pd.date_range(start='2023/1/5', end='2023/12/31', freq='ME')   # month end
pd.date_range(start='2023/1/5', end='2023/12/31', freq='MS')   # month start
pd.date_range(start='2023/1/5', end='2030/12/31', freq='YS')   # year start

pd.date_range(start='2023/1/5', periods=25, freq='D')          # count se

Common freq codes: D day, B business day, W week, h hour, min minute, s second, ME/MS month end/start, QE/QS quarter, YE/YS year.

4. `to_datetime()` ⭐⭐¶

String column ko datetime column mein convert karta hai. Real world mein sabse zyada use hota hai.

s = pd.Series(['2023/1/1', '2022/1/1', '2021/1/1'])
pd.to_datetime(s)

# Errors handle karna
pd.to_datetime(s, errors='coerce')       # galat values → NaT

# Format specify karna (fast + safe)
pd.to_datetime(s, format='%Y/%m/%d')

# read_csv ke waqt hi
pd.read_csv('data.csv', parse_dates=['date'])

5. `dt` accessor ⭐⭐¶

Ek datetime column se saari properties nikaalne ke liye.

df = pd.read_csv('messages.csv')
df['date'] = pd.to_datetime(df['date'])

df['date'].dt.year
df['date'].dt.month
df['date'].dt.month_name()
df['date'].dt.day
df['date'].dt.day_name()          # Monday, Tuesday...
df['date'].dt.dayofweek
df['date'].dt.hour
df['date'].dt.minute
df['date'].dt.second
df['date'].dt.quarter
df['date'].dt.is_month_start
df['date'].dt.is_month_end
df['date'].dt.is_quarter_start
df['date'].dt.is_year_start
df['date'].dt.is_leap_year
df['date'].dt.days_in_month
df['date'].dt.date                # sirf date part
df['date'].dt.time                # sirf time part

6. Real-world DateTime Examples ⭐¶

# Har mahine mein kitne messages
df['month_name'] = df['date'].dt.month_name()
df.groupby('month_name').size().plot(kind='bar')

# Har din ke messages
df['day_name'] = df['date'].dt.day_name()
df.groupby('day_name').size().plot(kind='bar')

# Weekend vs weekday
df['is_weekend'] = df['date'].dt.dayofweek.isin([5, 6])

# Saal ke hisaab se trend
df.groupby(df['date'].dt.year).size().plot()

# Stock data — sirf 2023 ka
stocks[stocks['date'].dt.year == 2023]

# Date range filter
mask = (df['date'] >= '2023-01-01') & (df['date'] <= '2023-06-30')
df[mask]

7. Timedelta — dates ka difference¶

d1 = pd.Timestamp('2023/1/1')
d2 = pd.Timestamp('2023/6/15')

d2 - d1                     # Timedelta('165 days')
(d2 - d1).days              # 165

pd.Timedelta(days=10)
d1 + pd.Timedelta(days=30)

# DataFrame mein
df['duration'] = df['end_date'] - df['start_date']
df['duration_days'] = df['duration'].dt.days

8. Time Series ke special features¶

# Date ko index bana do
df.set_index('date', inplace=True)

# Ab partial string indexing kaam karti hai
df['2023']              # poora 2023
df['2023-01']           # sirf January 2023
df['2023-01-05']        # ek din

# Resample ⭐ — time-based groupby
df['sales'].resample('ME').sum()      # monthly total
df['sales'].resample('W').mean()      # weekly average
df['sales'].resample('YE').sum()      # yearly

# Rolling window (moving average)
df['sales'].rolling(window=7).mean()  # 7-day moving average
df['sales'].expanding().mean()        # cumulative average

Advanced Pandas — Quick Revision¶

Kaam	Tool
Group-wise summary	`groupby().agg()`
Ek group nikaalna	`get_group()`
Group pe custom function	`groupby().apply()`
Tables jodna (stack)	`pd.concat()`
Tables jodna (SQL join)	`pd.merge()`, `df.join()`
Hierarchical index	`pd.MultiIndex`, `stack`, `unstack`, `swaplevel`
Wide → Long	`melt()`
Long → Wide + summary	`pivot_table()`
String operations	`.str.*`
Date operations	`pd.to_datetime()`, `.dt.*`, `date_range()`
Time-based grouping	`resample()`, `rolling()`

Advanced Pandas — Complete Notes¶

CampusX Session 19–22 (GroupBy + Merging/Joining/Concatenating + MultiIndex + Vectorized String Ops & DateTime + Pivot Table)¶

SESSION 19 — GROUPBY OBJECT IN PANDAS¶

1. GroupBy hai kya?¶

2. GroupBy object banana¶

3. Built-in Aggregation Functions¶

4. GroupBy Attributes & Methods¶

4.1 len() — kitne groups bane¶

4.2 size() — har group ka size¶

4.3 first(), last(), nth()¶

4.4 get_group() ⭐¶

4.5 groups attribute¶

4.6 describe()¶

4.7 sample()¶

4.8 nunique()¶

5. agg() — Multiple aggregations ek saath ⭐⭐¶

5.1 List pass karo¶

5.2 Dictionary pass karo (alag column, alag function)¶

5.3 Named aggregation (naye syntax)¶

6. GroupBy pe loop chalana¶

7. apply() on GroupBy ⭐¶

8. Multiple columns pe GroupBy ⭐¶

9. GroupBy — Practical Questions (Colab exercises)¶

---¶

SESSION 20 — MERGING, JOINING & CONCATENATING¶

1. pd.concat() ⭐¶

1.1 Vertically (axis=0, default) — rows add¶

1.2 Horizontally (axis=1) — columns add¶

1.3 keys — MultiIndex banana¶

1.4 Sirf common columns rakhna¶

2. pd.merge() ⭐⭐⭐¶

Syntax¶

2.1 Types of joins¶

2.2 Alag naam wale columns pe merge¶

2.3 Index pe merge¶

2.4 Suffixes (jab column names clash karein)¶

2.5 indicator¶

3. Relationship Types¶

4. df.join()¶

5. Merging — Practical Questions (Colab exercises)¶

---¶

SESSION 21 — MULTIINDEX SERIES & DATAFRAMES¶

1. MultiIndex kyun chahiye?¶

2. MultiIndex banana¶

2.1 from_tuples()¶

2.2 from_product() (aasan tareeqa)¶

3. MultiIndex Series¶

4. unstack() aur stack() ⭐¶

5. MultiIndex DataFrame¶

Columns pe bhi MultiIndex¶

6. MultiIndex DataFrame ke saath kaam¶

sort_index() on MultiIndex ⭐¶

swaplevel()¶

7. Long vs Wide Data ⭐⭐¶

8. melt() — Wide se Long ⭐¶

9. pivot_table() ⭐⭐⭐¶

Syntax¶

Plotting a pivot table ⭐¶

---¶

SESSION 22 — VECTORIZED STRING OPERATIONS & DATETIME¶

PART A — VECTORIZED STRING OPERATIONS (.str)¶

1. Problem kya hai?¶

2. Common .str methods¶

2.1 Case change¶

2.2 len()¶

2.3 strip() — whitespace hatana¶

2.4 split() aur get() ⭐¶

2.5 replace() ⭐¶

2.6 Filtering ke liye string methods ⭐¶

2.7 Check functions¶

2.8 Slicing¶

2.9 cat() — join karna¶

2.10 Padding¶

3. Regex ke saath .str ⭐¶

PART B — DATETIME IN PANDAS¶

1. Timestamp object¶

Attributes nikalna¶

Current time¶

2. DatetimeIndex¶

3. date_range() ⭐¶

4.1 `len()` — kitne groups bane¶

4.2 `size()` — har group ka size¶

4.3 `first()`, `last()`, `nth()`¶

4.4 `get_group()` ⭐¶

4.5 `groups` attribute¶

4.6 `describe()`¶

4.7 `sample()`¶

4.8 `nunique()`¶

5. `agg()` — Multiple aggregations ek saath ⭐⭐¶

7. `apply()` on GroupBy ⭐¶

1. `pd.concat()` ⭐¶

1.3 `keys` — MultiIndex banana¶

2. `pd.merge()` ⭐⭐⭐¶

2.5 `indicator`¶

4. `df.join()`¶

2.1 `from_tuples()`¶

2.2 `from_product()` (aasan tareeqa)¶

4. `unstack()` aur `stack()` ⭐¶

`sort_index()` on MultiIndex ⭐¶

`swaplevel()`¶

8. `melt()` — Wide se Long ⭐¶

9. `pivot_table()` ⭐⭐⭐¶

PART A — VECTORIZED STRING OPERATIONS (`.str`)¶

2. Common `.str` methods¶

2.2 `len()`¶

2.3 `strip()` — whitespace hatana¶

2.4 `split()` aur `get()` ⭐¶

2.5 `replace()` ⭐¶

2.9 `cat()` — join karna¶

3. Regex ke saath `.str` ⭐¶

1. `Timestamp` object¶

2. `DatetimeIndex`¶

3. `date_range()` ⭐¶

4. `to_datetime()` ⭐⭐¶

5. `dt` accessor ⭐⭐¶