Handling Categorical Data using Label Encoding
Categorical data like colors or categories, are convert into numerical values. Each category is assigned a unique code, enabling the computer to understand and process the information. For example, instead of using red or blue, we represent them with numbers, such as 1 for red and 2 for blue.
Label encoding involves converting categorical data into numerical format by assigning a distinct integer label to each category or class. In this encoding scheme, each unique category is mapped to an integer, making it easier for machine learning models to process and analyze the data.
Example: Code Implementation
Python3
from sklearn.preprocessing import LabelEncodercolors = ['red', 'blue', 'green', 'yellow', 'blue', 'green'] # samplelabel_encoder = LabelEncoder()encoded_colors = label_encoder.fit_transform(colors) # Fit and transform the dataprint("Original Colors:", colors)print("Encoded Colors:", encoded_colors) |
Output:
Original Colors: ['red', 'blue', 'green', 'yellow', 'blue', 'green']
Encoded Colors: [2 0 1 3 0 1]
Passing categorical data to Sklearn Decision Tree
Theoretically, decision trees are capable of handling numerical as well as categorical data, but, while implementing, we need to prepare the data for classification. There are two methods to handle the categorical data before training: one-hot encoding and label encoding. In this article, we understand how each method helps in converting categorical data and difference between both.