Patterns for Common Problems

Safe Division

Division by zero is a common source of production bugs. Your analysis works perfectly on test data, then crashes at 2 AM when one department has zero employees. Always protect division — it’s a habit that separates hobby code from production code.

# Dangerous: crashes if denominator is 0
ratio = numerator / denominator

# Safe: returns NaN instead of crashing
denominator = denominator.replace(0, float("nan"))
ratio = numerator / denominator

Pre-Aggregate to Avoid Slow Loops

If you find yourself writing a loop over groups, stop. .transform() or groupby-then-merge is almost always faster and cleaner.

# Slow: Python loop over departments
for dept in employee['department'].unique():
    dept_avg = employee[employee['department'] == dept]['salary'].mean()
        ...

# Fast: one-line transform
employee["dept_avg"] = (
    employee.groupby("department")["salary"].transform("mean")
)

Transform Then Filter

When a One-Liner is Better

Not everything needs multiple steps. Simple lookups are cleaner as one-liners:

# One-liner: perfectly readable
employee[employee["salary"] == employee["salary"].max()]

# Also works, but more code than needed: same result, more code
max_sal = employee["salary"].max()
employee[employee["salary"] == max_sal]

Use intermediate variables when: the logic has 3+ steps, you need to reuse a result, or the one-liner is hard to read.

Ratio Calculation

Highest Target Under Manager

Income By Title and Gender

Top Cool Votes

Key Takeaways

• Protect division: .replace(0, float("nan")) or if total > 0.
• Replace loops with .transform() or groupby-then-merge.
• Use intermediate variables for 3+ steps; one-liners for simple lookups.
• Every variable is a checkpoint for debugging.

What’s Next

You can now break complex analyses into clean, debuggable steps. Next: custom logic with .apply() — for when built-in methods can’t do what you need.