Classification algorithms classification-algorithms

This document provides an overview of various classification algorithms, focusing on their configuration, key parameters, and practical usage in advanced statistical models. Classification algorithms are used to assign categories to data points based on input features. Each section includes parameter descriptions and example code to help you implement and optimize these algorithms for tasks such as decision trees, random forest, and naive Bayes classification.

Decision Tree Classifier decision-tree-classifier

Decision Tree Classifier is a supervised learning approach used in statistics, data mining, and machine learning. In this approach, a decision tree is used as a predictive model for classification tasks, drawing conclusions from a set of observations.

Parameters

The table below outlines key parameters for configuring and optimizing the performance of a Decision Tree Classifier.

Example

Create MODEL modelname OPTIONS(
  type = 'decision_tree_classifier'
) AS
  select col1, col2, col3 from training-dataset

Factorization Machine Classifier factorization-machine-classifier

The Factorization Machine Classifier is a classification algorithm that supports normal gradient descent and the AdamW solver. The Factorization Machine classification model uses logistic loss, which can be optimized via gradient descent, and often includes regularization terms like L2 to prevent overfitting.

Parameters

The table below outlines key parameters for configuring and optimizing the performance of the Factorization Machine Classifier.

Example

CREATE MODEL modelname OPTIONS(
  type = 'factorization_machines_classifier'
) AS
  SELECT col1, col2, col3 FROM training-dataset

Gradient Boosted Tree Classifier gradient-boosted-tree-classifier

The Gradient Boosted Tree Classifier uses an ensemble of decision trees to improve the accuracy of classification tasks, combining multiple trees to enhance model performance.

Parameters

The table below outlines key parameters for configuring and optimizing the performance of the Gradient Boosted Tree Classifier.

Example

Create MODEL modelname OPTIONS(
  type = 'gradient_boosted_tree_classifier'
) AS
  select col1, col2, col3 from training-dataset

Linear Support Vector Classifier (LinearSVC) linear-support-vector-classifier

The Linear Support Vector Classifier (LinearSVC) constructs a hyperplane to classify data in a high-dimensional space. You can use it to maximize the margin between classes to minimize classification errors.

Parameters

The table below outlines key parameters for configuring and optimizing the performance of the Linear Support Vector Classifier (LinearSVC).

Example

Create MODEL modelname OPTIONS(
  type = 'linear_svc_classifier'
) AS
  select col1, col2, col3 from training-dataset

Logistic Regression logistic-regression

Logistic Regression is a supervised algorithm used for binary classification tasks. It models the probability that an instance belongs to a class using the logistic function and assigns the instance to the class with the higher probability. This makes it suitable for problems where the goal is to separate data into one of two categories.

Parameters

The table below outlines key parameters for configuring and optimizing the performance of Logistic Regression.

Example

Create MODEL modelname OPTIONS(
  type = 'logistic_reg'
) AS
  select col1, col2, col3 from training-dataset

Multilayer Perceptron Classifier multilayer-perceptron-classifier

The Multilayer Perceptron Classifier (MLPC) is a feedforward artificial neural network classifier. It consists of multiple fully connected layers of nodes, where each node applies a weighted linear combination of inputs, followed by an activation function. MLPC is used for complex classification tasks requiring non-linear decision boundaries.

Parameters

Example

CREATE MODEL modelname OPTIONS(
  type = 'multilayer_perceptron_classifier'
) AS
  select col1, col2, col3 from training-dataset

Naive Bayes Classifier naive-bayes-classifier

Naive Bayes Classifier is a simple probabilistic, multiclass classifier based on Bayes’ theorem with strong (naive) independence assumptions between features. It trains efficiently by computing conditional probabilities in a single pass over the training data to compute the conditional probability distribution of each feature given each label. For predictions, it uses Bayes’ theorem to compute the conditional probability distribution of each label given an observation.

Parameters

Example

CREATE MODEL modelname OPTIONS(
  type = 'naive_bayes_classifier'
) AS
  SELECT col1, col2, col3 FROM training-dataset

Random Forest Classifier random-forest-classifier

Random Forest Classifier is an ensemble learning algorithm that builds multiple decision trees during training. It mitigates overfitting by averaging predictions and selecting the class chosen by the majority of trees for classification tasks.

Parameters

Example

Create MODEL modelname OPTIONS(
  type = 'random_forest_classifier'
) AS
  select col1, col2, col3 from training-dataset

Next steps

After reading this document, you now know how to configure and use various classification algorithms. Next, refer to the documents on regression and clustering to learn about other types of advanced statistical models.