0.4. Data modeling

This section overviews all the scripts in the Chapter 4: Data modeling folder.

Definitions

Term	Definition
features	descriptive numerical representations to describe an object.
machine learning	the process of teaching a machine something that is useful.
classification model	If the goal is to separate out into classes (e.g. male or female), then this is known as a classification problem.
regression model	if the end goal is to measure some correlation with a variable and the output is more a numerical range (e.g. often between 0 and 1), then this is more of a regression problem.
deep learning models	models that are trained using a neural network.
unsupervised learning	f machines do not require labels (e.g. just need features), this is known as a unsupervised learning problem.
supervised learning	if machines require labels (e.g. male or female as separate feature arrays), this is known as a supervised learning problem.
training set	Machines are fed training data in the form of feature arrays and compress patterns in these feature arrays into models through algorithms.
testing set	data that is left out during training so that the accuracy can be calculated using cross-validation techniques.
validation set	data that is left out during training to tune hyperparameters (often used in deep learning modeling.
label	the tag of an audio sample (e.g. male or female) to aid in supervised learning.
cross-validation	how the performance of ML models are assessed (in terms of accuracy).

4.2 - Obtaining training data

Sample text

4.3 - Labeling training data

Sample text

4.4 - Classification models

Sample text

4.5 - Regression models

Sample text

4.6 - Deep learning models

Sample text

4.7 - AutoML approaches

Sample text