Welcome to the course! Cluster analysis is one of the types of unsupervised learning - an algorithm that works with unlabeled data (i.e. the data with no 'response' variable). Unlike Classification problems, there we don't exactly know if there is a clear relation between characteristics or how many groups are in a data. The main goal of unsupervised learning is to find 'hidden' structures or relations in data.

Before digging into different algorithms, you always need to perform an EDA (Exploratory Data Analysis). It includes anomaly detection (such as NaN or outliers), cleaning and preprocessing the data (detecting for missing values, or inappropriate formats), and some visualization to describe the simplest characteristics. Usually, the last part includes building box plots or bee swarm plots, or histograms.

Since our goal here is to divide the observations into groups, we mostly will use scatter plots using the seaborn library. If you hear that name for the first time, I highly recommend you to pass the Introduction course on seaborn. Let's start our analysis!