STATISTICA Data Miner课件.ppt

Agenda Introduction to STATISTICA Date Miner (SDM) Association Rules in SDM Clustering in SDM Classification Trees in SDM Short History of Statistics and Data Mining 1960s~1970s Parametric Methods and Linear Relationship In the 60s~70s centuries, games of chance were popular amount the wealthy, promoting many questions about probability to be addressed to famous mathematicians. These questions led to much research in mathematics and statistics during the ensuing years. The Rise of Modern Statistical Analysis(The Second Generation) 1980s Nonlinear Relationship in large data sets Multiple curvilinear regression Logit Model (including Logistic Regression) Probit Model (including Poisson Regression) The Generalized Linear Model (GLM) Machine Learning Methods (The Third Generation) Artificial Neural Networks Express a nonlinear function directly by means of assigning weights to the input variables, accumulate their effects and react to produce an output value following some sort of decision function Decision Trees Concerned with expressing the effects directly by developing methods to find rules that could be evaluated for separating the input values into one of several bins without having to express the functional relationship directly. Statistical Learning Theory (The Fourth Generation) Vector Space Meet the requirement/problem for nonlinear relationship and non-parametric (including the semi-parametric) Searching the Kernels to map Support Vector Machine (SVM) K-nearest neighbor (KNN) Bayesian Classifiers CRISP Cross-Industry Standard Process for Data Mining, was found in mid-1990s SEMMA SEMMA Process SAS? Sample Explore Modify Model Assess SDM-throughout Exploration Cleaning and Transformation Subsets and Feature Selection EDA (Exploratory Data Analysis) and graphical tools to uncover the potential patterns Modeling and Validation Deployment Data Miner Workspace How DMR does DMR provides a systematic method for building advanced analytic models to relate one o