Antonio Gulli - A collection of Data Science Interview Questions Solved in Python and Spark: Hands-on Big Data and Machine Learning

In his famous essay Computing Machinery and Intelligence Alan Turing asked a fundamental question " Can machines do what we (as thinking entities) can do? " Machine learning is not about thinking but more about a related activity: Learning or better, according to Arthur Samuel, the " Field of study that gives computers the ability to learn without being explicitly programmed ".

Machine learning techniques are typically classified into two categories:

In supervised learning pairs of examples made up by (inputs, desired output) are available and the computer learns a model according to which given an input, a desired output with a minimal error is predicted. Classification, Neural Networks and Regression are all examples of supervised learning. For all techniques we assume that there is an oracle or a teacher that can teach to computers what to do in order for them to apply the learned lessons on new unseen data.

In unsupervised learning computers have no teachers and they are left alone in searching for structures, patterns and anomalies in data. Clustering and Density Estimations are typical examples of unsupervised machine learning.

Let us now review the main machine learning techniques:

In classification the teacher presents pairs of (inputs, target classes) and the computer learns to attribute classes to new unseen data. Nave Bayesian, SVM, Decision Trees and Neural Networks are all classification methodologies. The first two are discussed in this volume, while the remaining ones will be part of the next volume.

In Regression the teacher presents pairs of (inputs, continuous targets) and computers learn how to predict continuous values on new and unseen data. Linear and Logistic regression are examples which will be discussed in the present volume. Decision Trees, SVM and Neural Networks can also be used for Regression.

In Associative rule learning c omputers are presented with a large set of observations, all being made up of multiple variables. The task is then to learn relations between variables such us A & B C (if A and B happen, then C will also happen).

In Clustering computers learn how to partition observations in various subsets, so that each partition will be made up of similar observations according to some well-defined metric. Algorithms like K-Means and DBSCAN belong also to this class.

In Density estimation computers learn how to find statistical values that describe data. Algorithms like Expectation Maximization belong also to this class.