JMP For: Analytical Application Development • Business Visualization • Design of Experiments • Exploratory Data Analysis • Interactive Data Mining • Modeling • Quality Improvement • Reliability • Statistics • Visual Six Sigma
JMP® for Interactive Data Mining
The JMP family of products offers both exploratory and predictive data mining techniques. With interactive data mining, you can find patterns in data with a large number of columns, rows or both. Using predictive data mining techniques, you can build models that make predictions that generalize well.
While JMP has classical techniques like regression and clustering, it also includes capabilities for building decision trees or regression trees and for creating neural network models. It’s easy to split your data to train, validate and test your candidate models, while the visualization tools enable you to easily review and compare them in a single, interactive environment.
Multithreaded code and data compression mean that it’s only hardware, and not JMP, that limits the scale of the problems you can tackle. Data is everywhere, and JMP’s ease of use makes mining data more accessible than ever before.
- Partition
- Neural
- Clustering
The Partition platform in JMP enables you to find cuts or groupings of inputs (Xs) that can best predict the variation in an output (Y). This process of splitting the data is recursive – you continue it until you get a useful fit. Grow your tree using decision trees, bootstrap forests (JMP Pro only) or boosted trees (JMP Pro only).
You can interactively build a decision tree with training and validation data.
The Neural platform in JMP enables you to build fully connected neural networks with hidden nodes in one or two layers. Each node can have one of three different activation functions, and you can have any number of nodes in each layer. Use boosting to help your network to learn difficult cases, and specify one of four penalty methods for your fit.
Compare the effect of different neural architectures on a boosted neural fit.
A key technique in unsupervised learning, clustering forms subgroups so that cases in a particular subgroup are more alike than those in another subgroup. The Cluster platform in JMP lets you scale and transform variables before analysis, and includes interactive hierarchical and K-means clustering.
Hierarchical clustering builds an interactive dendrogram that you can zoom in on to focus on interesting clusters. Heat maps and parallel coordinates plots help you to judge the optimal number of clusters to use.
More resources for Interactive Data Mining
Demos
On-Demand Webcasts
Slide Presentations
Successful Data Mining in Practice
Podcasts
Contact JMP Sales
877.594.6567 (US)
