Machine Learning From Zero To ...
Notes are taken from machinelearningmastery websit.
Machine Learning Key Ingredients:
- Data (training, test, validation)
- Training data is used to defining the features (or parameters?)
- Test data would not be used until model is complete, and then we test the model with it
- Model (Linear regression, Logistic regression, Neural network, Support vector machine, etc.)
- Loss function
- It determines how good the model is working
- Optimization method
Supervised Learning:
Supervised learning technique uses labeled dataset to train (supervise) algorithms into classifying data or predicting outcomes accurately.
- Classification:
- Creating a predictive mapping function from input variables (x) to discrete output variables (y)
- If there are two classes (types or categories of output) it’s called a binary classification
- If there a re more than two classes it’s called multi-class classification
- Uses an algorithm to accurately assign test data into specific categories, such as separating apples from oranges.
- Or, in the real world, supervised learning algorithms can be used to classify spam in a separate folder from your inbox.
- Linear classifiers, support vector machines, decision trees and random forest are all common types of classification algorithms
- Creating a predictive mapping function from input variables (x) to discrete output variables (y)
- Regression:
- Another type of supervised learning method that uses an algorithm to understand the relationship between dependent and independent variables.
- It is about creating a predictive mapping function from input variables (x) to continuous output variables (y)
- Regression models are helpful for predicting numerical values based on different data points, such as sales revenue projections for a given business.
- Some popular regression algorithms are linear regression, logistic regression and polynomial regression.
- Classification vs Regression:
- Sometimes there’s an overlap between classification and regression
- A classification algorithm may predict a continuous value, but the continuous value is in the form of a probability for a class label.
- A regression algorithm may predict a discrete value, but the discrete value in the form of an integer quantity.
- However, the way we evaluate classification and regression does not overlap and is unique:
- Classification predictions can be evaluated using accuracy, whereas regression predictions cannot.
- Regression predictions can be evaluated using root mean squared error, whereas classification predictions cannot.
- Sometimes there’s an overlap between classification and regression