Date and time feature extraction in ML Python - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

Extracting the hour from a datetime column

What is the output of this code that extracts the hour from a datetime column in a pandas DataFrame?

ML Python

import pandas as pd

df = pd.DataFrame({'timestamp': pd.to_datetime(['2024-06-01 08:15:27', '2024-06-01 22:45:00'])})
df['hour'] = df['timestamp'].dt.hour
print(df['hour'].tolist())

A [8, 22]

B [15, 45]

C [2024, 2024]

D [1, 1]

💡 Hint

The .dt accessor helps extract parts of datetime like hour, minute, or day.

The .dt.hour extracts the hour part from each datetime entry. The first timestamp has hour 8, the second 22.

🧠 Conceptual

intermediate

Why extract date and time features?

Why do we extract features like day of week or month from datetime data in machine learning?

A To reduce the size of the dataset by removing datetime columns.

B To convert datetime into text for natural language processing.

C To help the model learn patterns related to time, like weekends or seasons.

D To encrypt the datetime data for security.

💡 Hint

Think about how time affects behavior or events in real life.

Extracting features like day of week or month helps models find patterns related to time, such as weekend effects or seasonal trends.

❓ Metrics

advanced

Evaluating model with time-based features

You added 'day_of_week' and 'hour' features extracted from timestamps to your model. Which metric best shows if these features improved the model's ability to predict hourly sales?

A Mean Absolute Error (MAE) on a test set split randomly

B Accuracy score on a classification task

C Confusion matrix on training data

D Mean Squared Error (MSE) on a test set split by time (e.g., last month)

💡 Hint

Think about how to fairly test time-related predictions.

Using MSE on a test set split by time ensures the model generalizes to future unseen time periods, which is important for time-based features.

🔧 Debug

advanced

Fixing incorrect extraction of month from datetime

What error does this code raise when trying to extract the month from a datetime column?

ML Python

import pandas as pd

df = pd.DataFrame({'date': ['2024-06-01', '2024-07-15']})
df['month'] = df['date'].dt.month
print(df['month'])

A AttributeError: Can only use .dt accessor with datetimelike values

B KeyError: 'month' not found

C TypeError: unsupported operand type(s) for -: 'str' and 'int'

D No error, prints [6, 7]

💡 Hint

Check the data type of the 'date' column before using .dt accessor.

The 'date' column is string type, not datetime. The .dt accessor only works on datetime types, so it raises AttributeError.

❓ Model Choice

expert

Choosing model type for datetime feature importance

You have extracted multiple datetime features (hour, day_of_week, month) for a sales forecasting task. Which model type is best to capture complex interactions between these features and other variables?

A Linear Regression without interaction terms

B Decision Tree or Random Forest models

C K-Nearest Neighbors (KNN) without feature scaling

D Naive Bayes classifier

💡 Hint

Consider models that handle non-linear relationships and feature interactions well.

Decision Trees and Random Forests can capture complex, non-linear interactions between datetime features and other variables without needing manual feature engineering.