Interview Snippets
Overview
Common pandas questions asked in data science and analytics interviews. Each snippet includes the problem, solution, and explanation.
Data Manipulation
Find Duplicate Rows
Question: Find all duplicate rows in a DataFrame.
Find duplicates
df = pd.DataFrame({
'name': ['Alice', 'Bob', 'Alice', 'Charlie'],
'age': [25, 30, 25, 35]
})
# Find duplicates
duplicates = df[df.duplicated(keep=False)]
# name age
# 0 Alice 25
# 2 Alice 25
# Find first duplicate only
duplicates = df[df.duplicated(keep='first')]
# name age
# 2 Alice 25
# Count duplicates per combination
dup_counts = df.groupby(df.columns.tolist()).size().reset_index(name='count')
dup_counts = dup_counts[dup_counts['count'] > 1]
Top N per Group
Question: Get top 3 highest values per category.
Top N per group
df = pd.DataFrame({
'category': ['A', 'A', 'A', 'B', 'B', 'B', 'A', 'B'],
'value': [10, 20, 15, 25, 30, 22, 18, 28]
})
# Method 1: Using groupby + nlargest
top3 = df.groupby('category').apply(
lambda x: x.nlargest(3, 'value')
).reset_index(drop=True)
# Method 2: Using sort + groupby + head
top3 = (df.sort_values('value', ascending=False)
.groupby('category')
.head(3))
# Method 3: Using rank
df['rank'] = df.groupby('category')['value'].rank(ascending=False)
top3 = df[df['rank'] <= 3]
Second Highest Value
Question: Find the second highest salary per department.
Second highest value
df = pd.DataFrame({
'department': ['Sales', 'Sales', 'IT', 'IT', 'Sales'],
'salary': [50000, 60000, 70000, 65000, 55000]
})
# Method 1: Using nlargest
second_highest = df.groupby('department')['salary'].nlargest(2).groupby('department').last()
# Method 2: Using rank
df['rank'] = df.groupby('department')['salary'].rank(ascending=False, method='dense')
second_highest = df[df['rank'] == 2]
# Method 3: Using sort
second_highest = (df.sort_values('salary', ascending=False)
.groupby('department')
.nth(1))
Pivot with Aggregation
Question: Create a pivot table showing average sales by product and region.
Pivot table with aggregation
df = pd.DataFrame({
'product': ['A', 'B', 'A', 'B', 'A', 'B'],
'region': ['East', 'East', 'West', 'West', 'East', 'West'],
'sales': [100, 150, 120, 180, 110, 170]
})
# Pivot table
result = df.pivot_table(
index='product',
columns='region',
values='sales',
aggfunc='mean'
)
# region East West
# product
# A 105.0 120
# B 150.0 175
# With multiple aggregations
result = df.pivot_table(
index='product',
columns='region',
values='sales',
aggfunc=['mean', 'sum', 'count']
)
String Operations
Extract from String
Question: Extract domain from email addresses.
Extract domain from email
df = pd.DataFrame({
'email': ['alice@gmail.com', 'bob@yahoo.com', 'charlie@gmail.com']
})
# Extract domain
df['domain'] = df['email'].str.split('@').str[1]
# Or using extract
df['domain'] = df['email'].str.extract(r'@(.+)')
# email domain
# 0 alice@gmail.com gmail.com
# 1 bob@yahoo.com yahoo.com
# 2 charlie@gmail.com gmail.com
Clean Phone Numbers
Question: Standardize phone numbers to XXX-XXX-XXXX format.
Standardize phone numbers
df = pd.DataFrame({
'phone': ['(123) 456-7890', '555.123.4567', '9998887777']
})
# Remove all non-digits
df['cleaned'] = df['phone'].str.replace(r'\D', '', regex=True)
# Format as XXX-XXX-XXXX
df['formatted'] = df['cleaned'].str.replace(
r'(\d{3})(\d{3})(\d{4})',
r'\1-\2-\3',
regex=True
)
# phone cleaned formatted
# 0 (123) 456-7890 1234567890 123-456-7890
# 1 555.123.4567 5551234567 555-123-4567
# 2 9998887777 9998887777 999-888-7777
Missing Data
Fill Missing with Group Mean
Question: Fill missing values with the mean of their group.
Fill NaN with group mean
df = pd.DataFrame({
'category': ['A', 'A', 'B', 'B', 'A'],
'value': [10, np.nan, 15, 25, 20]
})
# Fill with group mean
df['value'] = df.groupby('category')['value'].transform(
lambda x: x.fillna(x.mean())
)
# category value
# 0 A 10.0
# 1 A 15.0 # Filled with mean of A: (10+20)/2
# 2 B 15.0
# 3 B 25.0
# 4 A 20.0
Forward Fill by Group
Question: Forward fill missing values within each group.
Forward fill within groups
df = pd.DataFrame({
'id': [1, 1, 1, 2, 2, 2],
'value': [10, np.nan, np.nan, 20, np.nan, 30]
})
# Forward fill per group
df['filled'] = df.groupby('id')['value'].ffill()
# id value filled
# 0 1 10.0 10.0
# 1 1 NaN 10.0
# 2 1 NaN 10.0
# 3 2 20.0 20.0
# 4 2 NaN 20.0
# 5 2 30.0 30.0
DateTime Operations
Calculate Age from Birthdate
Question: Calculate age from birthdate.
Calculate age
df = pd.DataFrame({
'name': ['Alice', 'Bob', 'Charlie'],
'birthdate': pd.to_datetime(['1990-05-15', '1985-08-20', '1995-12-10'])
})
today = pd.Timestamp('2024-01-15')
df['age'] = ((today - df['birthdate']).dt.days / 365.25).astype(int)
# name birthdate age
# 0 Alice 1990-05-15 33
# 1 Bob 1985-08-20 38
# 2 Charlie 1995-12-10 28
Business Days Between Dates
Question: Calculate business days between two dates.
Count business days
df = pd.DataFrame({
'start': pd.to_datetime(['2024-01-01', '2024-02-01']),
'end': pd.to_datetime(['2024-01-15', '2024-02-15'])
})
# Count business days
df['business_days'] = df.apply(
lambda row: len(pd.bdate_range(row['start'], row['end'])),
axis=1
)
# start end business_days
# 0 2024-01-01 2024-01-15 11
# 1 2024-02-01 2024-02-15 11
Filter Last N Days
Question: Get records from the last 30 days.
Filter recent records
df = pd.DataFrame({
'date': pd.date_range('2024-01-01', periods=100),
'value': range(100)
})
# Last 30 days from most recent date
end_date = df['date'].max()
start_date = end_date - pd.Timedelta(days=30)
recent = df[df['date'] >= start_date]
# Alternative
recent = df[df['date'] >= df['date'].max() - pd.Timedelta(days=30)]
Aggregation & GroupBy
Running Total by Group
Question: Calculate cumulative sum per category.
Cumulative sum per group
df = pd.DataFrame({
'category': ['A', 'A', 'A', 'B', 'B', 'B'],
'value': [10, 20, 15, 25, 30, 22]
})
df['cumsum'] = df.groupby('category')['value'].cumsum()
# category value cumsum
# 0 A 10 10
# 1 A 20 30
# 2 A 15 45
# 3 B 25 25
# 4 B 30 55
# 5 B 22 77
Percentage of Total per Group
Question: Calculate each value as percentage of group total.
Percentage within groups
df = pd.DataFrame({
'category': ['A', 'A', 'B', 'B'],
'value': [100, 150, 200, 250]
})
df['pct_of_group'] = df.groupby('category')['value'].transform(
lambda x: x / x.sum() * 100
)
# category value pct_of_group
# 0 A 100 40.0
# 1 A 150 60.0
# 2 B 200 44.44
# 3 B 250 55.56
Rank Within Groups
Question: Rank values within each group.
Rank within groups
df = pd.DataFrame({
'department': ['Sales', 'Sales', 'IT', 'IT', 'Sales'],
'salary': [50000, 60000, 70000, 65000, 55000]
})
df['rank'] = df.groupby('department')['salary'].rank(ascending=False)
# department salary rank
# 0 Sales 50000 3.0
# 1 Sales 60000 1.0
# 2 IT 70000 1.0
# 3 IT 65000 2.0
# 4 Sales 55000 2.0
Merging & Joining
Find Unmatched Records
Question: Find records in left DataFrame that don't have a match in right.
Find unmatched records
left = pd.DataFrame({
'id': [1, 2, 3, 4],
'name': ['Alice', 'Bob', 'Charlie', 'David']
})
right = pd.DataFrame({
'id': [1, 2, 5],
'value': [100, 200, 300]
})
# Find unmatched
merged = pd.merge(left, right, on='id', how='left', indicator=True)
unmatched = merged[merged['_merge'] == 'left_only']
# id name value _merge
# 2 3 Charlie NaN left_only
# 3 4 David NaN left_only
Merge Multiple DataFrames
Question: Merge three DataFrames on a common key.
Merge multiple DataFrames
df1 = pd.DataFrame({'id': [1, 2], 'a': [10, 20]})
df2 = pd.DataFrame({'id': [1, 2], 'b': [30, 40]})
df3 = pd.DataFrame({'id': [1, 2], 'c': [50, 60]})
# Method 1: Chain merges
result = df1.merge(df2, on='id').merge(df3, on='id')
# Method 2: Using reduce
from functools import reduce
dfs = [df1, df2, df3]
result = reduce(lambda left, right: pd.merge(left, right, on='id'), dfs)
# id a b c
# 0 1 10 30 50
# 1 2 20 40 60
Performance & Optimization
Optimize Data Types
Question: Reduce memory usage of a DataFrame.
Optimize memory usage
df = pd.DataFrame({
'category': ['A', 'B', 'A', 'B'] * 1000,
'value': range(4000)
})
# Check memory
print(df.memory_usage(deep=True))
# Optimize
df['category'] = df['category'].astype('category')
df['value'] = pd.to_numeric(df['value'], downcast='integer')
# Memory reduced significantly
print(df.memory_usage(deep=True))
Vectorized vs Loop
Question: Calculate a new column efficiently.
Vectorization vs loops
df = pd.DataFrame({
'a': range(10000),
'b': range(10000)
})
# Bad: Loop (slow)
result = []
for i in range(len(df)):
result.append(df.loc[i, 'a'] + df.loc[i, 'b'])
df['sum'] = result
# Good: Vectorized (fast)
df['sum'] = df['a'] + df['b']
# Bad: Apply with simple operation
df['sum'] = df.apply(lambda row: row['a'] + row['b'], axis=1)
# Good: Vectorized
df['sum'] = df['a'] + df['b']
Complex Queries
Consecutive Days
Question: Find records with 3 or more consecutive days of data.
Find consecutive sequences
df = pd.DataFrame({
'date': pd.to_datetime(['2024-01-01', '2024-01-02', '2024-01-03',
'2024-01-05', '2024-01-06', '2024-01-10'])
})
# Calculate day differences
df['day_diff'] = df['date'].diff().dt.days
# Mark new sequences
df['new_seq'] = (df['day_diff'] != 1) | (df['day_diff'].isna())
df['seq_id'] = df['new_seq'].cumsum()
# Count sequence lengths
seq_lengths = df.groupby('seq_id').size()
# Filter sequences with 3+ days
long_sequences = seq_lengths[seq_lengths >= 3]
result = df[df['seq_id'].isin(long_sequences.index)]
Remove Outliers by Group
Question: Remove outliers using IQR method per category.
Remove outliers per group
df = pd.DataFrame({
'category': ['A'] * 10 + ['B'] * 10,
'value': [1, 2, 3, 4, 5, 100, 7, 8, 9, 10, # 100 is outlier in A
20, 22, 24, 26, 28, 30, 32, 34, 200, 38] # 200 is outlier in B
})
def remove_outliers(group):
Q1 = group['value'].quantile(0.25)
Q3 = group['value'].quantile(0.75)
IQR = Q3 - Q1
lower = Q1 - 1.5 * IQR
upper = Q3 + 1.5 * IQR
return group[(group['value'] >= lower) & (group['value'] <= upper)]
cleaned = df.groupby('category').apply(remove_outliers).reset_index(drop=True)
Transpose with Aggregation
Question: Reshape data from long to wide with multiple value columns.
Complex pivot
df = pd.DataFrame({
'date': ['2024-01-01', '2024-01-01', '2024-01-02', '2024-01-02'],
'metric': ['sales', 'profit', 'sales', 'profit'],
'value': [100, 20, 120, 25]
})
# Pivot to wide format
wide = df.pivot(index='date', columns='metric', values='value')
# metric profit sales
# date
# 2024-01-01 20 100
# 2024-01-02 25 120
# Calculate margin
wide['margin'] = (wide['profit'] / wide['sales'] * 100).round(2)
SQL-Like Operations
Window Functions
Question: Calculate difference from previous row per group.
Lag/Lead calculations
df = pd.DataFrame({
'id': [1, 1, 1, 2, 2, 2],
'date': pd.date_range('2024-01-01', periods=6),
'value': [10, 15, 20, 30, 35, 40]
})
# Lag (previous value)
df['prev_value'] = df.groupby('id')['value'].shift(1)
# Calculate change
df['change'] = df['value'] - df['prev_value']
# Percentage change
df['pct_change'] = df.groupby('id')['value'].pct_change() * 100
Self Join
Question: Find pairs of records that meet a condition.
Self join pattern
df = pd.DataFrame({
'employee': ['Alice', 'Bob', 'Charlie'],
'manager': ['Bob', 'Charlie', None]
})
# Find employee-manager pairs
result = pd.merge(
df,
df,
left_on='manager',
right_on='employee',
suffixes=('_emp', '_mgr')
)
# employee_emp manager_emp employee_mgr manager_mgr
# 0 Alice Bob Bob Charlie
# 1 Bob Charlie Charlie None
Quick Tips
Common mistakes to avoid:
# Don't: Chained indexing
df[df['col'] > 5]['new_col'] = 10 # Warning!
# Do: Use loc
df.loc[df['col'] > 5, 'new_col'] = 10
# Don't: Loop through rows
for i in range(len(df)):
df.loc[i, 'new'] = df.loc[i, 'a'] + df.loc[i, 'b']
# Do: Vectorize
df['new'] = df['a'] + df['b']
# Don't: Repeated concatenation
result = pd.DataFrame()
for chunk in chunks:
result = pd.concat([result, chunk])
# Do: Collect then concat once
results = []
for chunk in chunks:
results.append(chunk)
result = pd.concat(results)
One-liners:
# Top 3 per group
df.sort_values('val', ascending=False).groupby('cat').head(3)
# Remove duplicates keeping last
df.drop_duplicates(subset=['id'], keep='last')
# Fill forward then backward
df['col'].ffill().bfill()
# Conditional column
df['label'] = np.where(df['val'] > 100, 'high', 'low')
# Multiple conditions
df['label'] = np.select(
[df['val'] < 50, df['val'] < 100],
['low', 'medium'],
default='high'
)
# Percentage change
df['pct_change'] = df['value'].pct_change() * 100
# Rank with ties
df['rank'] = df['score'].rank(method='dense', ascending=False)
# Group and get first/last
df.groupby('id').first()
df.groupby('id').last()
Interview preparation checklist:
- Know difference between loc and iloc
- Understand merge types (inner, left, right, outer)
- Master groupby operations
- Handle missing data appropriately
- Use vectorized operations over loops
- Know when to use apply vs vectorization
- Understand datetime operations
- Practice reshaping (pivot, melt)
- Optimize data types for memory
- Write readable, efficient code