DTREG 先進預測發展軟體

DTREG 先進預測發展軟體方法論包括:

Artificial neural networks (ANN)+ Support Vector Machine (SVM，支持向量機) + decision tree +K-Means Clustering（集群分析法）+ Linear Discriminant Analysis+Linear Regression models +Logistic Regression models+ Gene Expression Programming（基因表達規劃法） and Symbolic Regression + Time Series Analysis and Forecasting (Enterprise version only)

DTREG includes Correlation, Factor Analysis, Principal Components Analysis, and PCA Transformations of variables. DTREG also can perform time series analysis and forecasting

Features of Support Vector Machine (SVM) models:

?   SVM is a modern outgrowth of artificial neural networks. Highly accurate models. Classification and Regression analyses. Automatic grid search and pattern search for optimal parameters. Model building performance. Continuous, categorical and non-numeric variables. Missing value substitution. V-fold cross validation.

Features of Neural Network models:

?   Supports 3- and 4-layer perceptron network models, Radial Basis Function (RBF) neural networks,  self-organizing Cascade Correlation neural networks, GMDH polynomial networks, Probabilistic neural networks and General Regression neural networks.

?   Probabilistic neural networks are extremely accurate and fast to train.

Features of Time Series Analysis and Forecasting:

?   Choice of many types of base models including neural networks.

?   Automatic generation of lag, moving average, slope and trend variables.

?   Intervention variables

?   Automatic trend removal and variance stabilization

?   Autocorrelation calculation

?   Validation using hold-out rows at the end of the series

?   Several charts showing actual, validation, predicted, trend and residual values.