Data Mining and Predictive Modeling

Classification Trees (Partition)

Predict a categorical response as a function of predictor variables using recursive partitioning.

JMP features demonstrated:

Analyze > Modeling > Partition

Regression Trees (Partition)

Predict a continuous response as a function of predictor variables using recursive partitioning.

JMP features demonstrated:

Analyze > Modeling > Partition

Naive Bayes

Predict a categorical response as a function of predictor variables using Bayes conditional probabilities.

JMP features demonstrated:

Analyze > Modeling > Naive Bayes

K Nearest Neighbors

Predict a categorical or continuous response as a function of predictor variables using the outcomes of similar observations.

JMP features demonstrated:

Analyze > Modeling > K Nearest Neighbors

Neural Networks

Model complex relationships between inputs and outputs using flexible neural network models.

JMP features demonstrated:

Analyze > Modeling > Neural

Support Vector Machines

Modeling technique to predict (classify) binary and multi-class outcomes as a function of continous and/or categorial predictor variables

JMP features demonstrated:

Analyze > Predictive Modeling > Support Vector Machines

Creating a Validation Column (Holdout Sample)

This page describes how to create a validation column in JMP®. Validation, or out-of-sample cross-validation, is used to assess the predictive ability of a model.

JMP features demonstrated:

New Column, Initialize Data, Random Indicator, Value Labels

Text Explorer - Describing Unstructured Text Data

This page describes how to use the text explorer platform to describe the counts and frequencies of words and phrases in unstructured data.

JMP features demonstrated:

Term and phrase lists and word clouds.

Text Explorer - Analyzing Unstructured Text Data

This page describes how to use the text explorer platform to analyze unstructured text data in JMP and JMP Pro.

JMP features demonstrated:

Latent class analysis, latent semantic analysis, SVD scatterplots, and saving results.

Association Analysis (Market Basket Analysis)

Develop rules that indicate the likely occurrence of items based on the occurrence of other items to identify items that often appear together or identify dependent or associated events. 

JMP features demonstrated:

How to use association platform, which displays frequent item sets and rules by default.  Help with interpreting the output is provided.

Model Comparison and Selection

Use the Model Comparison platform to compare competing statistical models and select the best performing model.

JMP features demonstrated:

Analyze > Predictive Modeling > Model Comparison (also applies to the Model Comparison performed within the Formula Depot)