Encoding Categorical Variables
Categorical variables are features in your data that represent categories rather than numerical values. Examples include colors such as "red", "green", and "blue", or labels such as "yes" and "no". Machine learning algorithms require input data to be numeric, so you must convert these categorical variables into a numerical format before using them in models. This process is called encoding, and it ensures that your algorithms can interpret and learn from the data effectively.
One-hot encoding creates a new binary column for each category of a categorical variable. Each observation receives a 1 in the column corresponding to its category and 0 elsewhere.
12345678910111213import seaborn as sns import pandas as pd # Load Titanic dataset data = sns.load_dataset("titanic") # One-hot encode the 'embarked' column embarked_encoded = pd.get_dummies(data["embarked"], prefix="embarked") # Concatenate with original dataset data = pd.concat([data, embarked_encoded], axis=1) print(data[["embarked", "embarked_C", "embarked_Q", "embarked_S"]].head())
Label encoding assigns each unique category in a variable an integer value, transforming text labels into numbers.
123456789101112import seaborn as sns import pandas as pd from sklearn.preprocessing import LabelEncoder # Load Titanic dataset data = sns.load_dataset("titanic") # Label encode the 'sex' column encoder = LabelEncoder() data["sex_encoded"] = encoder.fit_transform(data["sex"]) print(data[["sex", "sex_encoded"]].head())
Order encoding assigns ordered integer values to categories based on their natural ranking. This method preserves the inherent order in ordinal categorical variables, such as education levels ("high school", "bachelor", "master", "doctorate").
123456789101112131415161718import seaborn as sns import pandas as pd # Load Titanic dataset data = sns.load_dataset("titanic") # Define the order of passenger classes: First < Second < Third class_order = ["First", "Second", "Third"] # Apply ordered categorical encoding data["class_encoded"] = pd.Categorical( data["class"], categories=class_order, ordered=True ).codes + 1 # +1 to make classes start from 1 instead of 0 # Display sample output print(data[["class", "class_encoded"]].head())
Be careful when encoding categorical variables. One-hot encoding can introduce the dummy variable trap - a situation where features are highly correlated, which may confuse some models. To avoid this, you can drop one of the dummy columns. Label encoding imposes an ordinal relationship between categories, which might not be appropriate for nominal data.
Thanks for your feedback!
Ask AI
Ask AI
Ask anything or try one of the suggested questions to begin our chat
Awesome!
Completion rate improved to 8.33Encoding Categorical Variables
Swipe to show menu
Categorical variables are features in your data that represent categories rather than numerical values. Examples include colors such as "red", "green", and "blue", or labels such as "yes" and "no". Machine learning algorithms require input data to be numeric, so you must convert these categorical variables into a numerical format before using them in models. This process is called encoding, and it ensures that your algorithms can interpret and learn from the data effectively.
One-hot encoding creates a new binary column for each category of a categorical variable. Each observation receives a 1 in the column corresponding to its category and 0 elsewhere.
12345678910111213import seaborn as sns import pandas as pd # Load Titanic dataset data = sns.load_dataset("titanic") # One-hot encode the 'embarked' column embarked_encoded = pd.get_dummies(data["embarked"], prefix="embarked") # Concatenate with original dataset data = pd.concat([data, embarked_encoded], axis=1) print(data[["embarked", "embarked_C", "embarked_Q", "embarked_S"]].head())
Label encoding assigns each unique category in a variable an integer value, transforming text labels into numbers.
123456789101112import seaborn as sns import pandas as pd from sklearn.preprocessing import LabelEncoder # Load Titanic dataset data = sns.load_dataset("titanic") # Label encode the 'sex' column encoder = LabelEncoder() data["sex_encoded"] = encoder.fit_transform(data["sex"]) print(data[["sex", "sex_encoded"]].head())
Order encoding assigns ordered integer values to categories based on their natural ranking. This method preserves the inherent order in ordinal categorical variables, such as education levels ("high school", "bachelor", "master", "doctorate").
123456789101112131415161718import seaborn as sns import pandas as pd # Load Titanic dataset data = sns.load_dataset("titanic") # Define the order of passenger classes: First < Second < Third class_order = ["First", "Second", "Third"] # Apply ordered categorical encoding data["class_encoded"] = pd.Categorical( data["class"], categories=class_order, ordered=True ).codes + 1 # +1 to make classes start from 1 instead of 0 # Display sample output print(data[["class", "class_encoded"]].head())
Be careful when encoding categorical variables. One-hot encoding can introduce the dummy variable trap - a situation where features are highly correlated, which may confuse some models. To avoid this, you can drop one of the dummy columns. Label encoding imposes an ordinal relationship between categories, which might not be appropriate for nominal data.
Thanks for your feedback!
