Why data cleaning consumes most analysis time in Data Analysis Python - Performance Analysis

Time Complexity: Why data cleaning consumes most analysis time

O(n)

Understanding Time Complexity

Data cleaning often takes the longest in data analysis. We want to understand why it costs so much time as data size grows.

How does the time needed to clean data change when we get more data?

Scenario Under Consideration

Analyze the time complexity of the following code snippet.

import pandas as pd

def clean_data(df):
    df = df.drop_duplicates()
    df = df.fillna(method='ffill')
    df['column'] = df['column'].apply(lambda x: x.strip() if isinstance(x, str) else x)
    return df

This code removes duplicates, fills missing values, and cleans text in one column.

Identify Repeating Operations

Identify the loops, recursion, array traversals that repeat.

• Primary operation: Scanning all rows multiple times for duplicates, missing values, and text cleaning.
• How many times: Each operation goes through the entire dataset once or more.

How Execution Grows With Input

As the number of rows grows, each cleaning step takes longer because it checks every row.

Input Size (n)	Approx. Operations
10	About 30 operations (3 passes over 10 rows)
100	About 300 operations (3 passes over 100 rows)
1000	About 3000 operations (3 passes over 1000 rows)

Pattern observation: The time grows roughly in direct proportion to the number of rows.

Final Time Complexity

Time Complexity: O(n)

This means the cleaning time grows linearly with the amount of data.

Common Mistake

[X] Wrong: "Data cleaning time stays the same no matter how much data there is."

[OK] Correct: Cleaning checks every row, so more data means more work and more time.

Interview Connect

Understanding how data cleaning scales helps you explain why it takes so long and shows you can think about real data challenges clearly.

Self-Check

"What if we added nested loops to compare every row with every other row during cleaning? How would the time complexity change?"

Practice

1. Why does data cleaning usually take the most time in data analysis?

easy

A. Because data cleaning is only about deleting data

B. Because data cleaning requires writing complex machine learning models

C. Because real-world data often has missing or incorrect values that need fixing

D. Because data cleaning is about creating visualizations

Solution

1. Step 1: Understand real-world data quality

   Real-world data is often messy with missing or wrong values.

2. Step 2: Recognize cleaning tasks

   Cleaning fixes these issues to prepare data for analysis.

3. Final Answer:

   Because real-world data often has missing or incorrect values that need fixing -> Option C

4. Quick Check:

   Messy data = cleaning time [OK]

Hint: Messy data means more cleaning time [OK]

Common Mistakes:

• Thinking cleaning is about building models
• Confusing cleaning with visualization
• Assuming cleaning just deletes data

2. Which Python code snippet correctly replaces missing values in a DataFrame column 'age' with the mean age?

easy

A. df['age'].dropna().mean()

B. df['age'].replace('NaN', df['age'].mean())

C. df.fillna('age', df.mean())

D. df['age'].fillna(df['age'].mean(), inplace=True)

Solution

1. Step 1: Identify correct method to fill missing values

   The fillna() method replaces missing values; inplace=True updates the DataFrame directly.

2. Step 2: Check correct usage of mean for replacement

   df['age'].mean() calculates the mean of the 'age' column correctly.

3. Final Answer:

   df['age'].fillna(df['age'].mean(), inplace=True) -> Option D

4. Quick Check:

   fillna with mean = correct syntax [OK]

Hint: Use fillna with mean() and inplace=True [OK]

Common Mistakes:

• Using replace with string 'NaN' instead of np.nan
• Wrong method fillna arguments
• Dropping missing values instead of filling

3. Given this code snippet cleaning a DataFrame:

import pandas as pd

df = pd.DataFrame({'score': [10, None, 20, None, 30]})
df['score'] = df['score'].fillna(int(df['score'].median()))
print(df['score'].tolist())

What is the output list?

medium

A. [10, 20, 20, 20, 30]

B. [10, 15, 20, 15, 30]

C. [10, null, 20, null, 30]

D. [10, 20, 20, null, 30]

Solution

1. Step 1: Calculate median of non-missing scores

   Non-missing values are [10, 20, 30]. Median is 20.

2. Step 2: Replace null with median

   Missing values replaced by 20, so list becomes [10, 20, 20, 20, 30].

3. Final Answer:

   [10, 20, 20, 20, 30] -> Option A

4. Quick Check:

   Median fill replaces null with 20 [OK]

Hint: Median of [10,20,30] is 20; fill null with 20 [OK]

Common Mistakes:

• Using mean instead of median
• Not replacing null values
• Confusing null with string 'null'

4. This code tries to remove duplicate rows from a DataFrame but does not work as expected:

df.drop_duplicates()
print(len(df))

What is the main issue?

medium

A. drop_duplicates() does not remove duplicates without inplace=True or reassignment

B. drop_duplicates() cannot remove duplicates in pandas

C. len(df) counts duplicates, so it is always wrong

D. drop_duplicates() requires a column name argument

Solution

1. Step 1: Understand drop_duplicates behavior

   By default, drop_duplicates() returns a new DataFrame and does not change original unless inplace=True or reassigned.

2. Step 2: Check code effect on original DataFrame

   Since result is not saved or inplace=True used, original df remains unchanged, so length stays same.

3. Final Answer:

   drop_duplicates() does not remove duplicates without inplace=True or reassignment -> Option A

4. Quick Check:

   Use inplace=True or assign result [OK]

Hint: Use inplace=True or assign result to remove duplicates [OK]

Common Mistakes:

• Assuming drop_duplicates changes df without inplace
• Thinking len(df) excludes duplicates automatically
• Believing drop_duplicates needs column argument always

5. You have a dataset with columns ['name', 'age', 'income']. Some age values are missing, and some income values are negative due to data entry errors. Which sequence of cleaning steps is best to prepare data for analysis?

hard

A. Delete all rows with missing 'age', then fill negative 'income' with mean

B. Fill missing 'age' with median, then replace negative 'income' with zero

C. Fill missing 'age' with zero, then delete rows with negative 'income'

D. Replace negative 'income' with median, then fill missing 'age' with mean

Solution

1. Step 1: Handle missing 'age' values

   Filling missing 'age' with median preserves data without bias from zeros or deletion.

2. Step 2: Fix negative 'income' errors

   Replacing negative 'income' with zero corrects errors without deleting data.

3. Final Answer:

   Fill missing 'age' with median, then replace negative 'income' with zero -> Option B

4. Quick Check:

   Median fill + zero fix = best cleaning [OK]

Hint: Fill missing with median; fix errors with zero [OK]

Common Mistakes:

• Deleting rows unnecessarily
• Filling missing with zero causing bias
• Replacing errors with mean without fixing negatives