Sign in with Google
Feature Engineering Basics
Join our community on Telegram!
Join the biggest community of Pharma students and professionals.
Feature engineering is the process of creating, transforming, or selecting variables in a dataset to improve the performance of statistical models or machine learning algorithms. These variables, known as features, are the inputs used by models to make predictions or discover patterns.
Raw data is often not in a suitable form for analysis or modeling. Feature engineering helps convert raw data into meaningful and useful features that better represent the underlying patterns in the data. This process can significantly improve the accuracy and effectiveness of models.
Common feature engineering techniques include creating new variables, transforming existing variables, handling categorical data, and scaling numerical values.
In R, feature engineering is often performed using base functions or the dplyr package.
library(dplyr)One basic technique is creating new features from existing variables. For example, suppose we have a dataset containing the length and width of a rectangle, and we want to create a new feature representing the area.
data <- data.frame(
length = c(5, 7, 9),
width = c(2, 3, 4)
)
data <- data %>%
mutate(area = length * width)Another common technique is transforming variables. For example, applying a logarithmic transformation to reduce skewness in a numerical variable.
data <- data %>%
mutate(log_area = log(area))Categorical variables often need to be converted into numerical form for modeling. This process is known as encoding. In R, this can be done using the factor() function.
data$category <- factor(c("A", "B", "A"))Scaling is another important feature engineering step. It ensures that numerical variables are on a similar scale, which is important for many machine learning algorithms.
# Standardize a variable
data$scaled_area <- scale(data$area)Feature engineering is a critical step in the data analysis and machine learning pipeline. Well-designed features can greatly improve model performance and lead to more accurate and meaningful results.
