SAS^® Machine Learning Features

Model development with modern machine learning algorithms 

• Decision forests:
  • Distribution of independent training runs.
  • Supports intelligent hyperparameter autotuning of model parameters.
  • Generation of SAS ASTOREs for production scoring.
• Gradient boosting:
  • Automated iterative search for optimal partition of the data in relation to selected label variable.
  • Automated resampling of input data with adjusted weights based on residuals.
  • Automated generation of weighted average for final supervised model.
  • Supports binary, nominal and interval labels.
  • Ability to customize tree training with variety of options for numbers of trees to grow, splitting criteria to apply, depth of subtrees and compute resources.
  • Automated stopping criteria based on validation data scoring to avoid overfitting.
  • Generation of SAS ASTOREs for production scoring.
• Neural networks:
  • Automated intelligent tuning of parameter set to identify optimal model.
  • Supports modeling of count data.
  • Intelligent defaults for most neural network parameters.
  • Ability to customize neural networks architecture and weights.
  • Techniques include deep forward neural network (DNN), convolutional neural networks (CNNs), recurrent neural networks (RNNs) and autoencoders.
  • Ability to use an arbitrary number of hidden layers to support deep learning.
  • Automatic standardization of input and target variables.
  • Automatic out-of-bag validation for early stopping to avoid overfitting.
  • Supports intelligent hyperparameter autotuning of model parameters.
  • Generation of SAS ASTOREs for production scoring.
  • Segmentation model for deep learning.
  • Network development platform for mobile or IoT device.
  • Deep learning and biomedical imaging algorithm used together to quickly identify and visualize shapes.
  • Load native DICOM files.
  • End-to-end pipeline to process audio streams and analyze audio data directly from a MIC or from large audio file.
• Support vector machines:
  • Models binary target labels.
  • Supports linear and polynomial kernels for model training.
  • Ability to include continuous and categorical in/out features.
  • Automated scaling of input features.
  • Ability to apply the interior-point method and the active-set method.
  • Supports data partition for model validation.
  • Supports cross-validation for penalty selection.
  • Generation of SAS ASTOREs for production scoring.
• Factorization machines
  • Supports the development of recommender systems based on sparse matrices of user IDs and item ratings.
  • Ability to apply full pairwise-interaction tensor factorization.
  • Includes additional categorical and numerical input features for more accurate models.
  • Supercharge models with timestamps, demographic data and context information.
  • Supports warm restart (update models with new transactions without full retraining).
  • Generation of SAS ASTOREs for production scoring.
• Bayesian networks:
  • Learns different Bayesian network structures, including naive, tree-augmented naive (TAN), Bayesian network-augmented naive (BAN), parent-child Bayesian networks and Markov blanket.
  • Performs efficient variable selection through independence tests.
  • Selects the best model automatically from specified parameters.
  • Generation of SAS ASTOREs for production scoring.
• Dirichlet Gaussian mixture models (GMM):
  • Can execute clustering in parallel and is highly multithreaded.
  • Performs soft clustering, which provides not only the predicted cluster score but also the probability distribution over the clusters for each observation.
  • Learns the best number of clusters during the clustering process.
  • Uses a parallel variational Bayes (VB) method as the model inference method. This method approximates the (intractable) posterior distribution and then iteratively updates the model parameters until it reaches convergence.
• Semisupervised learning algorithm:
  • Highly distributed and multithreaded.
  • Returns the predicted labels for both the unlabeled data table and the labeled data table.
• T-distributed stochastic neighbor embedding (t-SNE):
  • Highly distributed and multithreaded.
  • Returns low-dimensional embeddings that are based on a parallel implementation of the t-SNE algorithm.

Latest statistical algorithms

• Clustering:
  • K-means, k-modes or k-prototypes clustering.
  • Parallel coordinate plots to interactively evaluate cluster membership.
  • Scatter plots of inputs with cluster profiles overlaid for small data sets and heat maps with cluster profiles overlaid for large data sets.
  • Detailed summary statistics.
  • Generate on-demand cluster ID as a new column.
  • Supports holdout data (training and validation) for model assessment.
• Decision trees:
  • Computes measures of variable importance.
  • Supports classification and regression trees.
  • Based on a modified C4.5 algorithm or cost-complexity pruning.
  • Interactively grow and prune a tree. Interactively train a subtree.
  • Set tree depth, max branch, leaf size, aggressiveness of tree pruning and more.
  • Use tree map displays to interactively navigate the tree structure.
  • Generate on-demand leaf ID, predicted values and residuals as new columns.
  • Supports holdout data (training and validation) for model assessment.
  • Supports pruning with holdout data.
  • Supports autotuning.
• Logistic regression:
  • Models for binary data with logit and probit link functions.
  • Influence statistics.
  • Variable selection, including iteration plot.
  • Supports forward, backward, stepwise and lasso variable selection.
  • Frequency and weight variables.
  • Residual diagnostics.
  • Summary table includes overall ANOVA, model dimensions, fit statistics, model ANOVA, Type III test and parameter estimates.
  • Generate on-demand predicted values and residuals as new columns.
  • Supports holdout data (training and validation) for model assessment.
• Linear regression:
  • Influence statistics.
  • Variable selection, including iteration plot.
  • Supports forward, backward, stepwise and lasso variable selection.
  • Frequency and weight variables.
  • Residual diagnostics.
  • Summary table includes overall ANOVA, model dimensions, fit statistics, model ANOVA, Type III test and parameter estimates.
  • Generate on-demand predicted values and residuals as new columns.
  • Supports holdout data (training and validation) for model assessment
• Generalized linear models:
  • Distributions supported include beta, normal, binary, exponential, gamma, geometric, Poisson, Tweedie, inverse Gaussian and negative binomial.
  • Supports forward, backward, stepwise and lasso variable selection.
  • Variable selection, including iteration plot.
  • Offset variable support.
  • Frequency and weight variables.
  • Residual diagnostics.
  • Summary table includes model summary, iteration history, fit statistics, Type III test table and parameter estimates.
  • Informative missing option for treatment of missing values on the predictor variable.
  • Generate on-demand predicted values and residuals as new columns.
  • Supports holdout data (training and validation) for model assessment.  
• Generalized additive models:
  • Distributions supported include normal, binary, gamma, Poisson, Tweedie, inverse Gaussian and negative binomial.
  • Supports one- and two-dimensional spline effects.
  • GCV, GACV and UBRE methods for selecting the smoothing effects.
  • Offset variable support.
  • Frequency and weight variables.
  • Residual diagnostics.
  • Summary table includes model summary, iteration history, fit statistics and parameter estimates.
  • Supports holdout data (training and validation) for model assessment.  
• Nonparametric logistic regression:
  • Models for binary data with logit, probit, log-log and c-log-log link functions.
  • Supports one- and two-dimensional spline effects.
  • GCV, GACV and UBRE methods for selecting the smoothing effects.
  • Offset variable support.
  • Frequency and weight variables.
  • Market basket analysis.
  • The new NLHS_RANGE option in the PROC MBANALYSIS statement enables you to specify the range of number of items in the left-hand side (LHS) of a rule.
  • The new NRHS_RANGE option in the PROC MBANALYSIS statement enables you to specify the range of number of items in the right-hand side (RHS) of a rule.
  • The new ANTECEDENTLIST= option in the PROC MBANALYSIS statement enables you to specify the regular expression strings to match in the antecedent (left-hand side) of a rule.
  • The new CONSEQUENTLIST= option in the PROC MBANALYSIS statement enables you to specify the regular expression strings to match in the consequent (right-hand side) of a rule.
  • The new SEPARATOR= option in the PROC MBANALYSIS statement enables you to specify the separator character in the antecedent (left-hand side) or consequent (right-hand side) of a rule.
  • The maximum limit of the ITEMS= option is set to 1,000.
  • The maximum number of rules that are generated per thread on each node is 1 million.
• k-NN (K Nearest Neighbor)
  • Highly distributed and multithreaded.
  • Returns k-nearest neighbors based on a parallel implementation of the k-NN search algorithm .

Analytical data preparation

• Distributed data management routines provided via code:
  • T-distributed stochastic neighbor embedding (t-SNE).
  • Feature binning.
  • High-performance imputation of missing values in features with user-specified values, mean, pseudo median and random value of non-missing values.
  • Feature dimension reduction.
  • Large-scale principal components analysis (PCA), including moving windows and robust PCA.
  • Unsupervised learning with cluster analysis and mixed variable clustering.
• Large-scale data exploration and summarization.
• Large-scale data profiling of input data sources.
• Sampling: Supports random and stratified sampling, oversampling for rare events and indicator variables for sampled records.

Integrated text analytics

• Supports 32 native languages out of the box:
  
  • English.
    
  • Arabic.
    
  • Chinese.
    
  • Croatian.
    
  • Czech.
    
  • Danish.
    
  • Dutch.
    
  • Farsi.
    
  • Finnish.
    
  • French.
    
  • German.
    
  • Greek.
    
  • Hebrew.
    
  • Hindi.
    
  • Hungarian.
    
  • Indonesian.
    
  • Italian.
    
  • Japanese.
    
  • Korean.
    
  • Norwegian.
    
  • Polish.
    
  • Portuguese.
    
  • Romanian.
    
  • Russian.
    
  • Slovak.
    
  • Slovenian.
    
  • Spanish.
    
  • Swedish.
    
  • Tagalog.
    
  • Turkish.
    
  • Thai.
    
  • Vietnamese.
• Automated parsing, tokenization, part-of-speech tagging and lemmatization.
• Predefined concepts extract common entities such as names, dates, currency values, measurements, people, places and more.
• Automated feature extraction with machine-generated topics (singular value decomposition and latent Dirichlet allocation).
• Supports machine learning and rules-based approaches within a single project.
• Automatic rule generation with the BoolRule.
• Classify documents more accurately with deep learning (recurrent neural networks).

Model assessment

• Automatically calculates supervised learning model performance statistics.
• Produces output statistics for interval and categorical targets.
• Creates lift table for interval and categorical target.
• Creates ROC table for categorical target.

Model scoring

• Automatically generates SAS DATA step code for model scoring.
• Applies scoring logic to training, holdout data and new data.

SAS^® procedures (PROCs) & CAS actions

• A programming interface (SAS^® Studio) allows IT or developers to access a CAS server, load and save data directly from a CAS server, and support local and remote processing on a CAS server.
• Python programmers or IT staff can access data and perform basic data manipulation against a CAS server or execute CAS actions using PROC CAS.