Salford Predictive Modeler

The Salford Predictive Modeler, developed by Minitab, LLC, is a machine learning and predictive analytics software suite that aims to provide accurate and efficient model development for predictive, descriptive, and analytical purposes. According to the vendor, this comprehensive suite is designed for companies of various sizes across industries such as data science, business analysis, statistics, financial services, and retail.

Key Features

CART Modeling Engine: The CART modeling engine offers an advanced classification tree that can be used for both classification and regression modeling. It includes features such as hotspot detection, variable importance measures, user-defined splits, and automation tools for model tuning and experimentation.

Random Forests Modeling Engine: The Random Forests modeling engine leverages multiple alternative analyses, randomization strategies, and ensemble learning techniques. It is suitable for classification, regression, and clustering tasks and offers features like outlier detection, proximity heat map, multi-dimensional scaling, advanced missing value imputation, and variable importance measures.

MARS Modeling Engine: The MARS modeling engine is ideal for users who prefer results in a form similar to traditional regression. It captures essential nonlinearities and interactions and provides a graphical understanding of variable impact, variable importance measures, and automation tools for model tuning and experimentation.

TreeNet Modeling Engine: The TreeNet modeling engine is a flexible and powerful data mining tool that consistently generates highly accurate models. It offers features such as partial dependency plots, regression and classification loss functions, differential lift modeling, and automation tools for model tuning and experimentation.

Automation: The Salford Predictive Modeler includes over 70 pre-packaged automation scenarios inspired by leading model analysts. It automates various aspects of model development, including building, exploration, and refinement processes. The automation tools relieve users from repetitive tasks and allow them to focus on the creative aspects of model development. The suite also provides automation of variable selection, interaction detection, and experimentation with different learning and testing partitions.

Open Minitab Worksheet (.MTW) Functionality: The Salford Predictive Modeler allows users to import and work with Minitab worksheets seamlessly. This integration provides access to additional statistical and data analysis capabilities, enhanced data visualization, and reporting options.

BASIC Programming Language: The SPM suite includes a built-in BASIC programming language, enabling users to customize and extend the functionality of the software. The BASIC programming language allows users to create scripts and automate various tasks within the SPM suite.

LOGIT Modeling: The LOGIT module in SPM is a powerful tool for logistic regression analysis. Users can build, evaluate, and validate logistic regression models, perform prediction and scoring on new data, and assess model quality and reliability using regression diagnostics provided by the module.

Data Binning: SPM offers automated data binning, which simplifies complex datasets by organizing continuous variables into discrete bins. Users can choose from different binning methods and customize the process according to their specific requirements, potentially improving model performance.