SAS Enterprise Miner: Data Mining and Predictive Modeling

Apply variable selection and transformation in SAS Enterprise Miner to support modeling with continuous and binary targets, using numeric and categorical inputs, and variable clustering, decision tree, and transform nodes.

Explore SAS enterprise miner by partitioning data, applying variable selection with AOF variables, and running a regression on a continuous target. Use R-squared criteria to pick the most influential inputs.

Explore the value and significance of R-square in regression for data mining and predictive modeling in SAS Enterprise Miner, using interval-scaled target, numeric inputs, and variable selection.

The lecture demonstrates SAS Enterprise Miner variable selection, narrowing to five relevant inputs that strongly predict the interval target, followed by building a regression model and examining fit statistics.

Compare chi square and R-squared criteria for variable selection with a binary target variable, using a bank email response example and binning of continuous inputs for logistic regression.

An overview of variable selection and binary target modeling in SAS Enterprise Miner, highlighting which variables are selected or rejected, their relative importance, and the built regression and nominal-scale analyses.

Explore variable selection in SAS Enterprise Miner by comparing best variables and cluster component strategies within a two-path flow, leading to logistic regression and model comparison.

Explore variable frequency tables and cluster analyses in SAS Enterprise Miner for data mining and predictive modeling, including variable clustering, dendrograms, cluster plots, and regression model results.

Update model comparison in SAS Enterprise Miner across two regression models using variable clustering and variable selection strategies, showing differences in statistics, mean predictions, and model overlays.

Build a branched decision tree and regression model in SAS Enterprise Miner, using variable clustering and selection, data partitioning, and leaf-node segmentation for model comparison.

Explore variable selection and fit statistics in SAS Enterprise Miner, compare regression and decision tree models, using variable selection, variable clustering, and model comparison across training, validation, and test data.

Explore transforming variables in SAS Enterprise Miner using the transform variables node, comparing transformation before versus after variable selection while building regression models and assessing results.

Explore SAS Enterprise Miner regression modeling, focusing on variable selection before and after transformation, and evaluating lift, gain, and R2 to relate inputs to output variables.

Update and transform variables from partitioned data, apply optimal binning, and merge results into a regression model, while reviewing SAS code and model evaluation metrics.

Explore the side-by-side performance and combination of decision trees, regression, and neural networks for binary and ordinal targets, using flow diagrams, SAS code, and model comparisons.

Set neural network properties in SAS Enterprise Miner by selecting a multilayer perceptron and hidden units, then run data partition and imputation, ending with binary logistic regression and SAS code.

Analyze a binary output variable using decision trees, neural networks, and regression in SAS Enterprise Miner, comparing training and validation results, scores, lifts, and generated SAS code.

Explore combining regression models, decision trees, and neural networks to predict a continuous target (loss frequency) for ordinal outcomes using the Los data dataset and a SAS workflow.

Analyze regression node outputs, including target variable, training data, and ridge-optimized parameters. Review fit statistics, effects plots, lift, gain, SAS code, decision tree, and neural network results.

Examine how neural networks and gradient boosting run within SAS Enterprise Miner, using iteration plots, fit statistics, SAS code, and model performance rankings.

Explore the subseries plot and variable importance while examining binary target performance, error metrics, score rankings, and roc in a decision tree versus gradient boosting model, with SAS code.

Create an ensemble diagram that combines logistic regression, decision tree, and neural networks with an ensemble node, and explore average, maximum, and voting methods.

Explore SAS Enterprise Miner for data mining and predictive modeling, mastering regression, decision trees, neural networks, ensemble methods, and interpreting lift, gain, and score distributions across training and validation data.

Explore building decision tree models in SAS Enterprise Miner to predict binary responses and continuous targets, using data partitioning, profit matrices, and node-based trees for regression and classification.

Learn to run and update a decision tree model in SAS Enterprise Miner, covering data partitioning, train validation test sets, output variables, and key fit statistics and visualizations.

Apply the decision tree model to an independent prospect data set using a score node, predicting the probability of response and comparing models with a SAS code utility.

Build and compare a regression and decision tree model to predict risk using loss frequency data in SAS Enterprise Miner, including scoring, data partitioning, and SAS code generation.

Explore how a regression decision tree in SAS Enterprise Miner generates outputs like leaf statistics, a treemap, and model comparison to align mean and predicted values.

Explore interactive decision trees in SAS Enterprise Miner by building and modifying trees step by step, using a process flow with input, data partition, and three models for comparison.

Highlight result node data partition and refined data splits in SAS Enterprise Miner, featuring train, validation, and test sets and an interactive decision tree.

Explore building and refining an interactive decision tree in SAS Enterprise Miner, splitting nodes, editing rules, and assigning branches with immediate visual feedback.

Explore building a decision tree from the root in SAS Enterprise Miner, using interactive splits, entropy-driven variable selection, and subtree assessment to identify an optimal seven-leaf model, then compare models.

Examine neural network models in SAS Enterprise Miner to predict response and risk, using auto insurance use cases and a two-layer network with one hidden layer and hyperbolic tangent activation.

Run a neural network in SAS Enterprise Miner, interpret a binary target with predictor variables, and evaluate model fit using misclassification rate, mean square error, ROC, and AIC.

Explore how a neural network learns by tracking weights history across iterations, identifying optimal weights at the validation peak, and scoring new data with SAS Enterprise Miner.

Analyze neural network final weights, iteration-based weight changes, and fit statistics for binary target variables, using predictor and decision variables and SAS code generated by Enterprise Miner.

Score neural networks with SAS enterprise miner, compare models, and interpret ROC charts, lift curves, and SAS code outputs to evaluate predictive performance.

Explore neural network iteration plots in SAS enterprise miner, examining metrics like average square error, mean square error, root mean square error, and weights history to evaluate predictions and profits.

Explore neural network modeling in SAS by running nodes, reviewing lift, gain, and response overlays, and analyzing weights, errors, and SAS code plots.

Run eight neural networks and compare their results to reveal how cumulative lift, error metrics, and plots vary across models.

Compare neural, auto neural, and regression nodes in SAS Enterprise Miner to build and evaluate multiple predictive models, set targets, weights, and data partitions.

Explore auto neural node results in SAS enterprise miner, including final weights, training iterations, fit statistics, SAS code, and model plots, plus cumulative lift and gain metrics.

Compare three models across train and validation sets, highlighting auto neural's superior average squared error and misclassification rate, with lift and profit metrics.

Edit the binary target variable, configure prior probabilities and decisions, then build and evaluate three neural networks with data partitioning, weights history, and SAS code output.

Examine how switching to the average squared error function reshapes neural network results, revealing weight history, misclassification rates, and diagnostic plots like lift and mean squared error.

Explore how SAS enterprise miner conducts data mining and predictive modeling, examining iteration plots, weights across iterations, auto neural node results, cumulative lift, and generated SAS code for predictions.

Learn how to run the mine regression node in SAS Enterprise Miner, review predicted values, fit statistics, and model comparisons, including ROC, lift, and profit measures.

Explore regression models for binary, ordinal, nominal unordered, and continuous targets in SAS Enterprise Miner. Apply regression node properties to real-world cases like mail campaign prediction.

Explore how the regression node in SAS Enterprise Miner builds models, visualizes effects with plots, outputs SAS code, and analyzes lift, gain, and cumulative response for binary and ordinal targets.

Run a regression node to model an ordinal target with three levels and a nominal loss frequency, examining training, validation, and test data and the SAS code.

Update a regression node to view outputs, statistics, and plots in SAS Enterprise Miner, including target variables, residuals, training and validation data, and generated SAS code.

Create a flow diagram in SAS Enterprise Miner, configure a test data set and data partition, and run backward selection for logistic regression with a logit link.