Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Dr. Brian Mac Namee (www.comp.dit.ie/bmacnamee) Business Systems Intelligence: 5. Classification 1 2 of 25 55 Acknowledgments These notes are based (heavily) on those provided by the authors to accompany “Data Mining: Concepts & Techniques” by Jiawei Han and Micheline Kamber Some slides are also based on trainer’s kits provided by More information about the book is available at: www-sal.cs.uiuc.edu/~hanj/bk2/ And information on SAS is available at: www.sas.com 3 of 25 55 Classification & Prediction Today we will look at: – What are classification & prediction? – Issues regarding classification and prediction – Classification techniques: • • • • • • • Case based reasoning (k-nearest neighbour algorithm) Decision tree induction Bayesian classification Neural networks Support vector machines (SVM) Classification based on association rule mining concepts Other classification methods – Prediction – Classification accuracy 4 of 25 55 Classification & Prediction Classification: – Predicts categorical class labels – Classifies data (constructs a model) based on the training set and the values (class labels) in a classifying attribute and uses it in classifying new data Prediction: – Models continuous-valued functions, i.e., predicts unknown or missing values Typical Applications – Credit approval – Target marketing – Medical diagnosis – Treatment effectiveness analysis 5 of 25 55 Classification: A Two-Step Process 1) Model construction: – Each tuple/sample is assumed to belong to a predefined class, as determined by the class label attribute – The set of tuples used for model construction is the training set – A model created for classification 6 of 25 55 Classification: A Two-Step Process (cont…) 2) Model usage: – Estimate accuracy of the model • All members of an independent test-set is tested using the model built • The known label of test sample is compared with the classified result from the model • Accuracy rate is the percentage of test set samples that are correctly classified by the model – If the accuracy is acceptable, the model is used to classify data tuples whose class labels are not known 7 of 25 55 Classification: Model Construction Classification Algorithm Training Set NAME Mike Mary Bill Jim Dave Anne RANK YEARS TENURED Assistant Prof 3 no Assistant Prof 7 yes Professor 2 yes Associate Prof 7 yes Assistant Prof 6 no Associate Prof 3 no Classification Model IF rank = ‘professor’ OR years > 6 THEN tenured = ‘yes’ 8 of 25 55 Classification: Using The Model In Prediction Classifier Testing Set Unseen Data (Jeff, Professor, 4) NAME Tom Merlisa George Joseph RANK YEARS TENURED Assistant Prof 2 no Associate Prof 7 no Professor 5 yes Assistant Prof 7 yes Tenured? Yes 9 of 25 55 Supervised Vs. Unsupervised Learning Supervised learning (classification) – Supervision: The training data (observations, measurements, etc.) are accompanied by labels indicating the class of the observations – New data is classified based on the training set Unsupervised learning (clustering) – The class labels of training data is unknown – Given a set of measurements, observations, etc. with the aim of establishing the existence of classes or clusters in the data 10 of 25 55 Issues Regarding Classification & Prediction: Data Preparation Data cleaning – Preprocess data in order to reduce noise and handle missing values Relevance analysis (feature selection) – Remove the irrelevant or redundant attributes Data transformation – Generalize and/or normalize data 11 of 25 55 Issues Regarding Classification & Prediction: Evaluating Classification Methods Predictive accuracy Speed and scalability – Time to construct the model – Time to use the model Robustness – Handling noise and missing values Scalability – Efficiency in disk-resident databases Interpretability – Understanding and insight provided by the model 12 of 25 55 Classification Techniques: Case Based Reasoning (The k-Nearest Neighbor Algorithm) Case based reasoning is a classification technique which uses prior examples (cases) to determine the classification of unknown cases The k-nearest neighbour (k-NN) algorithm is the simplest form of case based reasoning 13 of 25 55 The k-Nearest Neighbor Algorithm) All instances correspond to points in n-D space The nearest neighbours are defined in terms of Euclidean distance (or other appropriate measure) The target value can be discrete or real-valued For discrete targets, k-NN returns the most common value among the k training examples nearest to the query For real-valued targets, k-NN returns a combination (e.g. average) of the nearest neighbours’ target values 14 of 25 55 Nearest Neighbour Example Features Class Wave Size (ft) Wave Period (secs) Good Surf? 6 15 Yes 1 6 No 5 11 Yes 7 10 Yes 6 11 Yes 2 1 No 3 4 No 6 12 Yes 4 2 No 10 ? Query 10 15 of 25 55 Nearest Neighbour Example When a new case is to be classified: f1 – Calculate the distance from the new case to all training cases – Put the new case in the same class as its nearest neighbour Wave Size ? ? ? Wave Period f2 16 of 25 55 k-Nearest Neighbour Example What about when it’s too close to call? Use the k-nearest neighbour technique f1 Wave Size 2 1 neighbours vs. neighbour ? Wave Period f2 – Determine the k nearest neighbours to the query case – Put the new case into the same class as the majority of its nearest neighbours 17 of 25 55 Nearest Neighbour Distance Measures Any kind of measurement can be used to calculate the distance between cases The measurement most suitable will depend on the type of features in the problem Euclidean distance is the most used technique d n (t i 1 i qi ) 2 where n is the number of features, ti is the ith feature of the training case and qi is the ith feature of the query case 18 of 25 55 Summary Of Nearest Neighbour Classification Strengths – No training involved – lazy learning – New data can be added on the fly – Some explanation capabilities – Robust to noisy data by averaging k-nearest neighbors Weaknesses – Not the most powerful classification – Slow classification – Curse of dimensionality One of the easiest machine learning classification techniques to understand 19 of 25 55 Case-Based Reasoning Uses lazy evaluation and analysis of similar instances However, instances are not necessarily “points in a Euclidean space” Methodology – Instances represented by rich symbolic descriptions – Multiple retrieved cases may be combined – Tight coupling between case retrieval, knowledge-based reasoning, and problem solving Lots of active research issues 20 of 25 55 Classification Techniques: Decision Tree Induction Decision trees are the most widely used classification technique in data mining today Formulate problems into a tree composed of decision nodes (or branch nodes) and classification nodes (or leaf nodes) Problem is solved by navigating down the tree until we reach an appropriate leaf node The tricky bit is building the most efficient and powerful tree J. Ross Quinlan is a famed researcher in data mining and decision theory. He has done pioneering work in the area of decision trees, including inventing the ID3 and C4.5 algorithms. 21 of 25 55 Training Dataset Age Income Student CreditRating BuysComputer <=30 high no fair no <=30 high no excellent no 31 - 40 high no fair yes >40 medium no fair yes >40 low yes fair yes >40 low yes excellent no 31 - 40 low yes excellent yes <=30 medium no fair no <=30 low yes fair yes >40 medium yes fair yes <=30 medium yes excellent yes 31 - 40 medium no excellent yes 31 - 40 high yes fair yes >40 medium no excellent no 22 of 25 55 Resultant Decision Tree Age? <=30 30 - 40 Student? no No >40 Credit Rating? Yes yes excellent Yes No fair Yes 23 of 25 55 Algorithm For Decision Tree Induction Basic algorithm (a greedy algorithm) – Tree is constructed in a top-down recursive divide-and-conquer manner – At the start, all the training examples are at the root – Attributes are categorical (if continuous-valued, they are discretized in advance) – Examples are partitioned recursively based on selected attributes – Test attributes are selected on the basis of a heuristic or statistical measure (e.g. information gain) 24 of 25 55 Algorithm For Decision Tree Induction Conditions for stopping partitioning – All samples for a given node belong to the same class – There are no remaining attributes for further partitioning – majority voting is employed for classifying the leaf – There are no samples left 25 of 25 55 Attribute Selection Measure: Information Gain (ID3/C4.5) The attribute selection mechanism used in ID3 and based on work on information theory by Claude Shannon If our data is split into classes according to fractions {p1,p2…, pm} then the entropy is measured as the info required to classify any arbitrary tuple as follows: m E ( p1 ,p2 ,...,pm ) pi log 2 pi i 1 26 of 25 55 Attribute Selection Measure: Information Gain (ID3/C4.5) (cont…) The information measure is essentially the same as entropy At the root node the information is as follows: 9 5 info[9,5] E , 14 14 9 9 5 5 log 2 log 2 14 14 14 14 0.94 27 of 25 55 Attribute Selection Measure: Information Gain (ID3/C4.5) (cont…) To measure the information at a particular attribute we measure info for the various splits of that attribute 28 of 25 55 Attribute Selection Measure: Information Gain (ID3/C4.5) (cont…) At the age attribute the information is as follows: 5 4 5 info[2,3], [4,0], [3,2] info2,3 info4,0 info3,2 14 14 14 5 2 2 3 3 log 2 log 2 14 5 5 5 5 4 4 4 0 0 log 2 log 2 14 4 4 4 4 5 3 3 2 2 log 2 log 2 14 5 5 5 5 0.694 29 of 25 55 Attribute Selection Measure: Information Gain (ID3/C4.5) (cont…) In order to determine which attributes we should use at each node we measure the information gained in moving from one node to another and choose the one that gives us the most information 30 of 25 55 Attribute Selection By Information Gain Example Class P: BuysComputer = “yes” Class N: BuysComputer = “no” – I(p, n) = I(9, 5) =0.940 Compute the entropy for age: Age <=30 <=30 31 - 40 >40 >40 >40 31 - 40 <=30 <=30 >40 <=30 31 - 40 31 - 40 >40 Income high high high medium low low low medium low medium medium medium high medium Student no no no no yes yes yes no yes yes yes no yes no CreditRating fair excellent fair fair fair excellent excellent fair fair fair excellent excellent fair excellent BuysComputer no no yes yes yes no yes no yes yes yes yes yes no Age pi ni I(pi, ni) >=30 2 3 0.971 30 – 40 4 0 0 >40 3 2 0.971 31 of 25 55 Attribute Selection By Information Gain Computation 5 4 5 E (age) I (2,3) I (4,0) I (3,2) 14 14 14 0.694 5 I (2,3) means “age <=30” has 5 out of 14 samples, 14 with 2 yes and 3 no. Hence: Gain(age) I ( p, n) E (age) 0.246 Similarly: Gain(income) 0.029 Gain( student ) 0.151 Gain(credit _ rating ) 0.048 32 of 25 55 Other Attribute Selection Measures Gini index (CART, IBM IntelligentMiner) – All attributes are assumed continuous-valued – Assume there exist several possible split values for each attribute – May need other tools, such as clustering, to get the possible split values – Can be modified for categorical attributes 33 of 25 55 Extracting Classification Rules From Trees Represent knowledge in the form of IF-THEN rules One rule is created for each path from root to leaf Each attribute-value pair along a path forms a conjunction The leaf node holds the class prediction Rules are easier for humans to understand IF Age = “<=30” AND Student = “no” THEN BuysComputer = “no” IF Age = “<=30” AND Student = “yes” THEN BuysComputer = “yes” IF Age = “31…40” THEN BuysComputer = “yes” IF Age = “>40” AND CreditRating = “excellent” THEN BuysComputer = “yes” IF Age = “<=30” AND CreditRating = “fair” THEN BuysComputer = “no” 34 of 25 55 Overfitting Training Set Test Set 35 of 25 55 Overfitting (cont…) Training Set Test Set 36 of 25 55 Avoiding Overfitting In Classification An induced tree may overfit the training data – Too many branches, some may reflect anomalies due to noise or outliers – Poor accuracy for unseen samples Two approaches to avoiding overfitting – Prepruning: Halt tree construction early • Do not split a node if this would result in a measure of the usefullness of the tree falling below a threshold • Difficult to choose an appropriate threshold – Postpruning: Remove branches from a “fully grown” tree to give a sequence of progressively pruned trees • Use a set of data different from the training data to decide which is the “best pruned tree” 37 of 25 55 Approaches To Determine The Final Tree Size Separate training (2/3) and testing (1/3) sets Use cross validation, e.g., 10-fold cross validation Use all the data for training – But apply a statistical test (e.g., chi-square) to estimate whether expanding or pruning a node may improve the entire distribution Use minimum description length (MDL) principle – Halting growth of the tree when the encoding is minimized 38 of 25 55 Enhancements To Basic Decision Tree Induction Allow for continuous-valued attributes – Dynamically define new discrete-valued attributes that partition the continuous attribute value into a discrete set of intervals Handle missing attribute values – Assign the most common value of the attribute – Assign probability to each of the possible values Attribute construction – Create new attributes based on existing ones that are sparsely represented – This reduces fragmentation, repetition, and replication 39 of 25 55 Classification In Large Databases Classification - a classical problem extensively studied by statisticians and machine learning researchers Scalability: Classifying data sets with millions of examples and hundreds of attributes with reasonable speed Why decision tree induction in data mining? – Relatively faster learning speed (than other classification methods) – Convertible to simple and easy to understand classification rules – Can use SQL queries for accessing databases – Comparable classification accuracy with other methods 40 of 25 55 Data Cube-Based Decision-Tree Induction Integration of generalization with decision-tree induction Classification at primitive concept levels – E.g., precise temperature, humidity, outlook, etc. – Low-level concepts, scattered classes, bushy classification-trees – Semantic interpretation problems Cube-based multi-level classification – Relevance analysis at multi-levels – Information-gain analysis with dimension + level 41 of 25 55 Decision Tree In SAS 42 of 25 55 Bayesian Classification: Why? Probabilistic learning: – Calculate explicit probabilities for a hypothesis – Among the most practical approaches to certain types of learning problems Incremental: – Each training example can incrementally increase/ decrease the probability that a hypothesis is correct – Prior knowledge can be combined with observed data Probabilistic prediction: – Predict multiple hypotheses, weighted by their probabilities Standard: – Bayesian methods can provide a standard of optimal decision making against which other methods can be measured 43 of 25 55 Bayesian Theorem: Basics Let X be a data sample whose class label is unknown Let H be a hypothesis that X belongs to class C For classification problems, determine P(H|X): the probability that the hypothesis holds given the observed data sample X – P(H): prior probability of hypothesis H (i.e. the initial probability before we observe any data, reflects the background knowledge) – P(X): probability that sample data is observed – P(X|H): probability of observing the sample X, given that the hypothesis holds 44 of 25 55 Bayesian Theorem Given training data X, posteriori probability of a hypothesis H, P(H|X) follows the Bayes theorem P ( X | H ) P ( H ) P(H | X ) P( X ) Informally, this can be written as posterior = (likelihood * prior) / evidence MAP (maximum posteriori) hypothesis h arg max P(h | D) arg max P(D | h)P(h). MAP hH hH Practical difficulty: require initial knowledge of many probabilities, significant computational cost 45 of 25 55 Naïve Bayes Classifier A simplified assumption: attributes are conditionally independent: n P( X | C i) P( x k | C i ) k 1 The product of occurrence of say 2 elements x1 and x2, given the current class is C, is the product of the probabilities of each element taken separately, given the same class P([y1,y2],C) = P(y1,C) * P(y2,C) No dependence relation between attributes Greatly reduces the computation cost, only count the class distribution. Once the probability P(X|Ci) is known, assign X to the class with maximum P(X|Ci)*P(Ci) 46 of 25 55 age Class: <=30 C1:buys_computer= <=30 ‘yes’ 30…40 C2:buys_computer= >40 >40 ‘no’ >40 31…40 Data sample <=30 X =(age<=30, <=30 Income=medium, >40 Student=yes <=30 Credit_rating= 31…40 Fair) 31…40 >40 Training dataset income student credit_rating high no fair high no excellent high no fair medium no fair low yes fair low yes excellent low yes excellent medium no fair low yes fair medium yes fair medium yes excellent medium no excellent high yes fair medium no excellent buys_computer no no yes yes yes no yes no yes yes yes yes yes no 47 of 25 55 Naïve Bayesian Classifier: Example Compute P(X/Ci) for each class P(age=“<30” | buys_computer=“yes”) = 2/9=0.222 P(age=“<30” | buys_computer=“no”) = 3/5 =0.6 P(income=“medium” | buys_computer=“yes”)= 4/9 =0.444 P(income=“medium” | buys_computer=“no”) = 2/5 = 0.4 P(student=“yes” | buys_computer=“yes)= 6/9 =0.667 P(student=“yes” | buys_computer=“no”)= 1/5=0.2 P(credit_rating=“fair” | buys_computer=“yes”)=6/9=0.667 P(credit_rating=“fair” | buys_computer=“no”)=2/5=0.4 X=(age<=30 ,income =medium, student=yes,credit_rating=fair) P(X|Ci) : P(X|buys_computer=“yes”)= 0.222 x 0.444 x 0.667 x 0.0.667 =0.044 P(X|buys_computer=“no”)= 0.6 x 0.4 x 0.2 x 0.4 =0.019 P(X|Ci)*P(Ci ) : P(X|buys_computer=“yes”) * P(buys_computer=“yes”)=0.028 P(X|buys_computer=“no”) * P(buys_computer=“no”)=0.007 X belongs to class “buys_computer=yes” 48 of 25 55 Naïve Bayesian Classifier: Comments Advantages : – Easy to implement – Good results obtained in most of the cases Disadvantages – Assumption: class conditional independence , therefore loss of accuracy – Practically, dependencies exist among variables – E.g., hospitals: patients: Profile: age, family history etc – Symptoms: fever, cough etc., Disease: lung cancer, diabetes etc – Dependencies among these cannot be modeled by Naïve Bayesian Classifier How to deal with these dependencies? – Bayesian Belief Networks 49 of 25 55 Bayesian Networks Bayesian belief network allows a subset of the variables conditionally independent A graphical model of causal relationships – Represents dependency among the variables – Gives a specification of joint probability distribution Y X Z P • Nodes: random variables • Links: dependency • X,Y are the parents of Z, and Y is the parent of P • No dependency between Z and P • Has no loops or cycles 50 of 25 55 Bayesian Belief Network: An Example Family History Smoker (FH, S) LungCancer PositiveXRay Emphysema Dyspnea Bayesian Belief Networks (FH, ~S) (~FH, S) (~FH, ~S) LC 0.8 0.5 0.7 0.1 ~LC 0.2 0.5 0.3 0.9 The conditional probability table for the variable LungCancer: Shows the conditional probability for each possible combination of its parents n P( z1,..., zn ) P( z i | Parents( Z i )) i 1 51 of 25 55 Learning Bayesian Networks Several cases – Given both the network structure and all variables observable: learn only the CPTs – Network structure known, some hidden variables: method of gradient descent, analogous to neural network learning – Network structure unknown, all variables observable: search through the model space to reconstruct graph topology – Unknown structure, all hidden variables: no good algorithms known for this purpose D. Heckerman, Bayesian networks for data mining 52 of 25 55 Lazy Vs. Eager Learning Lazy learning: – Case based reasoning Eager learning: – Decision-tree and Bayesian classification Key differences: – Lazy method may consider query instance when deciding how to generalize beyond the training data D – Eager method cannot since they have already chosen global approximation when seeing the query 53 of 25 55 Lazy Vs. Eager Learning Efficiency: – Lazy, less time training but more time predicting Accuracy: – Lazy method effectively uses a richer hypothesis space since it uses many local linear functions to form its implicit global approximation to the target function – Eager learners must commit to a single hypothesis that covers the entire instance space – Easier for lazy learners to cope with concept drift 54 of 25 55 Summary Classification is an extensively studied problem Classification is probably one of the most widely used data mining techniques with a lot of extensions Classification techniques can be categorized as either lazy or eager Scalability is still an important issue for database applications: thus combining classification with database techniques should be a promising topic Research directions: classification of non-relational data, e.g., text, spatial, multimedia, etc. classification of skewed data sets 55 of 25 55 Questions?