The number of different variables (attributes), or types of data, that describe an individual element of data that a machine learning system will have to find patterns on to learn. For example, if the data describes a person, we might have three dimensions that we want the system to learn: age, income, and location. Each dimension adds greater value in terms of differentiation and pattern discovery but also adds to the complexity of the machine learning and need for greater amounts of data. The dimensions that we need all come down to what we want the machine to learn and be able to predict. You can use the data from dimensions you know to predict data in a dimension you’d like to know.