Refactor code and enforce robustness
Company: Capital One
Role: Data Scientist
Category: Coding & Algorithms
Difficulty: medium
Interview Round: Onsite
You are given this Python script:
# script.py
import pandas as pd
DATA_PATH = 'data.csv'
result = None
def compute_total(col):
df = pd.read_csv(DATA_PATH)
total = 0
for x in df[col]:
if x == '':
total += 0
else:
total += float(x)
print(total)
compute_total('amount')
Tasks: (1) Identify at least five defects or risks (correctness, performance, readability, resource management, security). (2) Refactor into a small, testable module with clear interfaces, type hints, and no global state; include input validation and assert-based precondition checks (and explain when assertions vs. exceptions are appropriate). (3) Write three pytest-style unit tests using assert statements that cover normal, missing/NaN, and malformed inputs. (4) Provide an environment.yml for a Conda environment (Python 3.11, pinned dependencies) and the exact commands to create/activate it. (5) Explain the benefits of modularization for maintainability, dependency management, and testability. (6) State the time/space complexity before and after refactoring and any I/O bottlenecks you’d address.
Quick Answer: This question evaluates proficiency in code refactoring, robustness, input validation, unit testing, environment specification, and complexity analysis for Python data-processing workflows (pandas), within the Coding & Algorithms domain for Data Scientist roles.
