Date and time feature extraction in ML Python - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is date and time feature extraction in machine learning?

It is the process of taking raw date and time data and turning it into useful pieces of information (features) that a machine learning model can understand and use to make predictions.

beginner

Name three common features extracted from a date-time value.

Year, month, day of the week, hour, minute, and second are common features extracted from date-time values.

beginner

Why might extracting the day of the week be useful for a model?

Because some patterns depend on the day, like sales being higher on weekends or traffic being heavier on weekdays.

intermediate

How can you extract the hour from a timestamp in Python using pandas?

Use pandas' dt accessor: df['timestamp'].dt.hour extracts the hour part from a datetime column.

intermediate

What is the benefit of extracting cyclical features like hour or month as sine and cosine values?

It helps the model understand that these features repeat in cycles, for example, hour 23 is close to hour 0, which normal numbers don't show well.

Which feature is NOT typically extracted from a datetime value?

A Day of the week

B Hour of the day

C User's age

D Month

User's age is not a direct feature from datetime values; it requires additional data.

Why do we sometimes convert hour or month into sine and cosine values?

A To capture the cyclical nature of time

B To make the data categorical

C To reduce the number of features

D To normalize the data between 0 and 1

Sine and cosine transformations help models understand that time features repeat in cycles.

Which Python library is commonly used to extract date and time features?

A NumPy

B pandas

C matplotlib

D scikit-learn

pandas has built-in datetime functions to extract features easily.

What does df['date'].dt.weekday return in pandas?

A The day of the week as an integer (Monday=0)

B The day of the month

C The week number in the year

D The month number

It returns the day of the week as an integer where Monday is 0 and Sunday is 6.

Which feature might help predict traffic patterns best?

A Year

B Microsecond

C Second

D Day of the week

Traffic often changes depending on the day of the week (weekdays vs weekends).

Explain how you would extract useful features from a timestamp for a machine learning model.

Think about breaking down the timestamp into parts and representing repeating patterns.

Why is it important to transform cyclical time features like hour or month before using them in models?

Consider how time repeats and how numbers alone might mislead the model.