Additional GPS Generalized Path Seeker Features
Additional MARS Features
Additional GPS Generalized Path Seeker Features
Additional Random Forests Features are available in Pro, ProEx, and Ultra.
Additional TreeNet Features
Advances in Gradient Boosting: The Power of Post Processing
Advances in Gradient Boosting: the Power of Post-ProcessingLearn how TreeNet stochastic gradient boosting can be improved by post processing techniques such as GPS Generalized Path Seeker, RuleLearner, and ISLE.
I. Gradient Boosting and Post-Processing:
- What is missing from Gradient Boosting?
- Why post-processing techniques are used?
II. Applications Benefiting from Post-Processing: Examples from a variety of industries.
- Financial Services
III. Typical Post-Processing Steps
- Generalized Path Seeker (GPS): Modern high-speed LASSO-style regularized regression
- Importance Sampled Learning Ensembles (ISLE): identify and reweight the most influential trees
- RuleLearner: ISLE on “steroids.” Identify the most influential nodes and rules
V. Case Study Example
- Output/Results without Post-Processing
- Output/Results with Post-Processing
Watch the Video
Components and Features
|SPM Components and Features||What's New|
|SPM Components and Features||What's New|
|CART (Classification and Regression Trees)||User defined linear combination lists for splitting; Constrains on trees; Automatic addition of missing value indicators; Enhanced GUI reporting; User controlled Cross Validation; Out-of-bag performance stats and predictions; Profiling terminals nodes based on user supplied variables; Comparison of Train vs. Test consistency across nodes; RandomForests-style variable importance|
|MARS (Automated Nonlinear Regression)||Updated GUI interface; Model performance based on independent test sample or Cross Validation; Support for time series models|
|TreeNet (Gradient Boosting, Boosted Trees)||One-Tree TreeNet (CART alternative); RandomForests via TreeNet (RandomForests regression alternative) Interaction Control Language (ICL); Interaction strength reporting; Enhanced partial dependency plots; RandomForests-style randomized splits;|
|RandomForests (Bagging Trees)||RandomForests regression; Saving out-of-bag scores; Speed enhancements|
|High-Dimensional Multivariate Pattern Discovery||Battery Target (link) to identify mutual dependencies in the data|
|Unsupervised Learning (Breiman's Column Scrambler)||New|
|Model Compression and Rule Extraction||New: ISLE; RuleLearner; Hybrid Compression|
|Automation||56 pre-packaged scenarios based on years of high-end consulting|
|Parallel Processing||New: Automatic support of multiple cores via multithreading|
|Hotspot Detection||Segment Extraction (Battery Priors)|
|Missing Value Handling and Imputation|
|Outlier Detection||New: GUI reports, tables, and graphs|
|Linear Methods for Regression, Recent Advances and Discoveries||New: OLS Regression; Regularized Regression Including: LAR/LASSO Regression; Ridge Regression; Elastic Net Regression/ Generalized Path Seeker|
|Linear Methods for Classification, Recent Advances and Discoveries||New: LOGIT; LAR/LASSO; Ridge; Elastic Net/ Generalized Path Seeker|
|Model Assessment and Selection||Unified reporting of various performance measures across different models|
|Ensemble Learning||New: Battery Bootstrap; Battery Model|
|Time Series Modeling||New|
|Model Simplification Methods|
|Data Preparation||New: Battery Bin for automatic binning of a user selected set of variables with large number of options|
|Large Data Handling||64 bit support; Large memory capacity limited only by your hardware|
|Model Translation (SAS, C, Java, PMML, Classic)||Java|
|Data Access (all popular statistical formats supported)||Updated Stat Transfer Drivers including R workspaces|
|Model Scoring||Score Ensemble (combines multiple models into a powerful predictive machine)|
Autodiscovery leverages the stability advantages of multiple trees to rank variables for importance and thus select a subset of predictors for modeling. In SPM® v8.0 and earlier Autodiscovery runs a very simple training data only TreeNet model growing out to 200 trees. The variable importance ranking generated from this model is then used to reduce the list of all available predictors down to the top performing predictors in this background model. Autodiscovery is fast and easy, as there are no control parameters to set, but it is just a mechanism for quickly testing whether a substantial refinement in the number of predictors can improve model performance.
Free Download: The SPM® Software Suite
General Features Introduction
SPM 6.6 (TreeNet TN 6.4) or greater supports data access to Microsoft SQL Server, Oracle, MySQL and other RDMS via ODBC interface.
Since SQL Queries cannot be entered via standard Windows ODBC dialog data source selection dialog, one has to use command line to open data directly from SQL Server.
SPM 8 Introduction
- Automated Non-Linear Regression
- MARS software is ideal for users who prefer results in a form similar to traditional regression while capturing essential nonlinearities and interactions. The MARS approach to regression modeling effectively uncovers important data patterns and relationships that are difficult, if not impossible, for other regression methods to reveal. MARS builds its model by piecing together a series of straight lines with each allowed its own slope. This permits MARS to trace out any pattern detected in the data.
- High-Quality Probability
- The MARS model is designed to predict continuous numeric outcomes such as the average monthly bill of a mobile phone customer or the amount that a shopper is expected to spend in a web site visit. MARS is also capable of producing high quality probability models for a yes/no outcome. MARS performs variable selection, variable transformation, interaction detection, and self-testing, all automatically and at high speed.
- High-Performance Results
- Areas where MARS has exhibited very high-performance results include forecasting electricity demand for power generating companies, relating customer satisfaction scores to the engineering specifications of products, and presence/absence modeling in geographical information systems (GIS).
SPM® 8 Product Versions
- The best of the best. For the modeler who must have access to leading edge technology available and fastest run times including major advances in ensemble modeling, interaction detection and automation. ULTRA also provides advance access to new features as they become available in frequent upgrades.
- For the modeler who needs cutting-edge data mining technology, including extensive automation of workflows typical for experienced data analysts and dozens of extensions to the Salford data mining engines.
- A true predictive modeling workbench designed for the professional data miner. Variety of supporting conventional statistical modeling tools, programming language, reporting services, and a modest selection of workflow automation options.
- Literally the basics. Salford Systems award winning data mining engines without extensions or automation or surrounding statistical services, programming language, and sophisticated reporting. Designed for small budgets while still delivering our world famous engines
Feature Matrix and download PDF
I have 6 years using SAS and other statistical software to conduct academic and business projects. I started using SPM to build predictive models in May 2014. Our team mainly uses SPM TreeNet to build models for direct mail campaigns. I think the SPM software (Salford Predictive Modeler) is S.P.M. - SMART, PRODUCTIVE and MANAGEABLE.
Fast with big data
We use a lot of data. Most of the time, our model data sets have hundreds of thousands records and thousands of variables. SPM can handle these large data sets super-fast and builds predictive models in as short as few minutes. It also gives out pop-up messages if it finds some data issues so that I can identify problems more easily.
Powerful Battery tools to reduce variables
When I build predictive models with thousands of variables, I find one of the hardest tasks is to reduce the number of important variables. My goal is move from over a 1,000 variables to fewer than 20.
Models with fewer important variables without losing much lift are much easier to implement in our business environment. SPM provides 31 powerful Battery tools to do this for me. The top 3 Battery options that I most often use are Shaving, LOVO and Keep. All three can help you remove those least contributed variables from the model in order to maintain those most important predictors in your model.
Machine learning for missing values
While I build TreeNet models in SPM, I don’t need to spend lots of time dealing with missing values because SPM can take care of this for me and automatically learns the pattern from the build data set and then assigns proper values for the missing records. This feature saves me lots of time and manual work.
User friendly interface and no programming required
SPM makes model building easy for me even though I’m not programmer or statistician. With its user friendly interface design, it is easy to build a robust model in a few minutes. No hard coding needed to build models. It saves me lots of time in programming and code testing.
Build not only good, but reliable models
SPM provides many algorithms like Cart, TreeNet, and Random Forests, which I can choose to build different models. For instance, instead of building one decision tree one time, TreeNet is able to build hundreds of decisions trees in minutes and find the optimal one for me.
The CVR battery tool also helps me validate the model performance by building 20 or 30 models with different cross validation sets. This has helped tremendously in improving reliability of our models when I have thin data.
Easy scoring even with millions records
The scoring feature from SPM makes scoring data super easy. If I need to score data that has a relatively smaller size (hundreds of thousands records), I can get it done with SPM on my Windows environment in a couple of minutes. If I need to score a large data set that has millions of customer records, I can export the model from SPM into SAS and then do the scoring on SAS server without any problem.
Manage modeling process
The statistics summary file helps track the modeling process, report error message and show descriptive statistics information to help me manage the modeling and scoring process.
Helpful support team
The support team from SPM has rich knowledge in model building and is very helpful when I have questions. They take my questions or requests by email or via phone calls and always get back me in a timely manner with helpful answer. Additionally, they provide useful resources that help me better understand the topics.
A user's license sets a limit on the amount of learn sample data that can be analyzed. The learn sample is the data used to build the model. Note that there is no limit to the number of test sample data points that may be analyzed. In other words, rows -by- columns of variable and observations used to build the model. Variable not used in the model do not count. Observations reserved for testing, or excluded for other reasons, do not count.