Athabasca University (AU ) • Data Mining

1 What types of inputs do each of the 3 types of decision trees take? ID3: categorical only C4.5: numeric and categorical CART: numeric and categorical 2 How do each of the 3 types of decision trees split? ID3: max information gain and minimum entropy C4.5: information gain and gain ratio CART: Gini index and twoing creation 3 How do each of the 3 types of decision trees handle missing data? ID3: does not C4.5: omits from calculations CART: surrogate splits 4 What is CBA? ...

1 Factors that influence the organization of requirements -nature and scope of the system under development; -Techniques and strategies used to elicit and capture reqts; -Tools and environments employed 2 Organizing by Subsystem Complex systems can only be visualized and built as systems of subsystems; 1st build system-level reqts wo/ref to subsystems; Refine into subsystems/define interfaces; Hierarchy of requirements 3 Organizing by Product Families Common functionality across...

1 What is the goal of software development? "-Develop quality software, -Meet schedule, -Meet budget, -Satisfy customer needs (but customers are varied)" 2 What is the relationship between requirements engineering and project success? "The Standish Group 1994: 1/3 of all projects cancelled and 1/2 overbudget, Reqts errors most common class of error, Reqts errors most expensive errors to fix (multiplicitive effect/rework), Better reqts engr => better project success" 3 S...

P(E) is assumed to be the same for all ___ ______ (Naive Bayes) Naive Bayes Pros - Despite strict independence assumptions, performs surprisingly well for classification on real-world tasks - Natural and incremental learner. Needs not reprocess all past training examples when new data arrive - Fast, efficient, and effective quantile(titanic$Age, seq(from=0, to =1, by = .2)) Create a quintile for titanic explaining the Age variable titanic_w1_c50 <- C5.0(Survived~.,titanic) Build a cla...

1 Precision (a or positive) or Positive Predictive Value (PPV) TP/(TP+FP) Positive Predictive Value (PPV) = TP/(TP+FP) 2 Precision (b or negative) or Negative Predictive Value (NPV) TN/(TN+FN) Negative Predictive Value (NPV) = TN/(TN+FN) 3 Recall (a or positive) = True Positive Rate (a) or TPR(a) Sensitivity =TP/(TP+FN) 4 Recall (b or negative) = True Negative Rate (b) Specificity = TN/(TN+FP) 5 F-MEASURE OR F-SCORE 3 items (2 bullet points and the formula) 1...

1 What types of inputs do each of the 3 types of decision trees take? ID3: categorical only C4.5: numeric and categorical CART: numeric and categorical 2 How do each of the 3 types of decision trees split? ID3: max information gain and minimum entropy C4.5: information gain and gain ratio CART: Gini index and twoing creation 3 How do each of the 3 types of decision trees handle missing data? ID3: does not C4.5: omits from calculations CART: surrogate splits 4 What is CBA? ...