This documentation site is for previous versions. Visit our new documentation site for current releases.

This content has been archived and is no longer being updated.

Links may not function; however, this content may be relevant to outdated versions of the product.

# Computation models

Updated on April 5, 2022

The process of model development allows you to create such default models as regression, decision tree, genetic algorithm, and bivariate.

## Regression models

Regression models work well on very linear data. The Prediction Studio logistic regression models are a generalization of linear regression models. They represent the predictive model as a formula where the various predictors are added up after multiplication by a coefficient, the resulting outcome being fit through a logistic function that maps the outcomes to a range between 0 to 1. The regression models can be viewed as the coefficients of the formula or as a scorecard.

## Decision tree models

Decision tree models work well on mid-volume, highly non-linear data. In a decision tree, the predictive model is represented in a tree-like structure, with conditions in each node. Predictors in each node consist of numeric values (in the case of numeric predictors) or list of values (in the case of symbolic predictors).

## Genetic algorithms

Genetic algorithms are an optimization method that is inspired by natural evolution. This method is used to obtain a non-linear, highly predictive model. Genetic algorithm is an iterative algorithm where each generation consists of a number of models. In the first generation, the models have a low average performance that improves in following generations while also maintaining diversity. When the performance has converged after N generations, the model with the highest performance in the last generation is saved.

## Bivariate models

Bivariate models add bivariate analysis to Prediction Studio. and model the relationship between all possible pairings of the predictors calculating the potential performance of each pair as if the relationship between them was perfectly modeled, identifying the best operators to model the relationship, calculating its predictive performance, as well as the percentage rating of the potential performance.

## Have a question? Get answers now.

Visit the Support Center to ask questions, engage in discussions, share ideas, and help others.

Did you find this content helpful?

Want to help us improve this content?