df.isnull().sum()         # count NaN per column
df.isnull().sum() / len(df) * 100  # percentage missing

df.count()           # non-null count per column ✅
df['col'].count()    # for single column

df.dropna()                        # drop rows with ANY NaN
df.dropna(subset=['review_text'])  # drop rows where specific column is NaN ✅
df.dropna(how='all')               # drop only if ALL values are NaN

df.fillna(0)                        # fill with constant
df['col'].fillna(df['col'].mean())  # fill with mean
df['col'].fillna(df['col'].median()) # fill with median
df.fillna(method='ffill')           # forward fill (time series)
df.fillna(method='bfill')           # backward fill

df['col'].nunique()      # number of unique values ✅
df['col'].unique()       # array of unique values
df['col'].value_counts() # count of each unique value

df.groupby('cuisine')['rating'].mean()   # average rating per cuisine ✅

df.groupby('col')['val'].mean()    # average
df.groupby('col')['val'].sum()     # total
df.groupby('col')['val'].count()   # count
df.groupby('col')['val'].max()     # maximum
df.groupby('col')['val'].min()     # minimum
df.groupby('col')['val'].median()  # median
df.groupby('col')['val'].std()     # standard deviation

df.drop_duplicates()                    # all columns
df.drop_duplicates(subset=['col'])      # specific column ✅
df['col'].drop_duplicates().values      # memory-efficient unique values ✅

df.corr()                              # correlation matrix for all columns
df['col1'].corr(df['col2'])            # correlation between two columns

# pivot_table + corr for customer similarity
df.pivot_table(
    index='customer_id',
    columns='product_category',
    values='purchase_amount'
).fillna(0).corr()                     # ✅ exam answer

df.sort_values('col')                   # ascending
df.sort_values('col', ascending=False)  # descending
df.sort_values(['col1', 'col2'])        # multiple columns

pd.merge(df1, df2, on='key')              # inner join (default)
pd.merge(df1, df2, on='key', how='left')  # left join
pd.merge(df1, df2, on='key', how='outer') # outer join
pd.concat([df1, df2])                     # stack vertically

df[df['rating'] < 3.0]                          # single condition ✅
df[(df['rating'] < 3) & (df['city'] == 'NYC')]  # multiple conditions
df[df['col'].isin(['val1', 'val2'])]            # filter by list
df[df['col'].str.contains('pattern')]           # string filter

df['col'].str.lower()           # lowercase
df['col'].str.upper()           # uppercase
df['col'].str.strip()           # remove whitespace
df['col'].str.replace('a', 'b') # replace
df['col'].str.split(',')        # split
df['col'].str.contains('pat')   # boolean mask

df.head()        # first 5 rows
df.tail()        # last 5 rows
df.info()        # column types + non-null counts
df.describe()    # summary statistics
df.shape         # (rows, columns)
df.dtypes        # data type of each column
df.columns       # column names

df.dtypes                          # check all types
df['col'].astype('int')           # convert to integer
df['col'].astype('float')         # convert to float
df['col'].astype('str')           # convert to string
df['col'].astype('category')      # convert to category (memory efficient)
pd.to_datetime(df['date_col'])    # convert to datetime

pd.read_csv('file.csv')
pd.read_json('file.json')
pd.read_parquet('file.parquet')
pd.read_excel('file.xlsx')
pd.read_sql(query, connection)

df.to_csv('output.csv', index=False)
df.to_json('output.json')
df.to_parquet('output.parquet')
df.to_excel('output.xlsx', index=False)

from collections import Counter

# Count frequencies
Counter(['a', 'b', 'a', 'c'])     # {'a': 2, 'b': 1, 'c': 1}
Counter.most_common(5)            # top 5 most frequent ✅

# Count URL segments in log analysis
Counter(
    url.split('/')[2]
    for url in urls
    if len(url.split('/')) > 2
)