Which methods are primarily used in Exploratory Data Analysis (EDA)?

In Exploratory Data Analysis (EDA), the primary methods used are univariate and bivariate analyses. Univariate analysis focuses on examining each variable individually to understand its distribution and characteristics. This can involve techniques such as histograms, box plots, and summary statistics that provide insights into central tendency, variability, and the shape of the data distribution.

Bivariate analysis, on the other hand, examines the relationship between two variables. This is crucial in understanding correlations, interactions, and dependencies as it can reveal trends and patterns that might not be apparent when looking at variables in isolation. Techniques such as scatterplots, correlation coefficients, and contingency tables are often used to visualize and analyze these relationships.

By employing both univariate and bivariate methods, a machine learning engineer can gain a comprehensive understanding of dataset characteristics and how different variables interact with each other. The combination of insights obtained from these two analyses is foundational for preprocessing data and informing the selection and application of appropriate machine learning models.