Knowledge Discovery
1 like623 views
Knowledge discovery involves non-trivially extracting useful and previously unknown information from data. It includes data mining to detect patterns in prepared data. Knowledge discovery can be divided into classification, numerical prediction, association, and clustering and is used for applications like financial forecasting, targeted marketing, medical diagnosis, and fraud detection. Data mining can use labeled data with known attributes to predict unknown attributes through supervised learning, or unlabeled data without known attributes through unsupervised learning techniques like association rules and clustering.
Knowledge Discovery
- 2. DefinitionDefinition – “Non-trivial extraction of implicit, previously unknown and potentially useful information from data.”Data Mining – Responsible for detecting patterns from the pre-processed (prepared) data. It is only a part of Knowledge discovery process.
- 3. ApplicationsCan be divided into four major kinds:ClassificationNumerical predictionAssociationClusteringSome examples:Automatic abstractionFinancial forecastingTargeted marketingMedical diagnosisCredit card fraud detectionWeather forecasting etc.
- 4. Labeled & Unlabeled dataGeneral Terminology: Instances – Dataset of examplesAttributes – Variables in an instanceLabeled dataSpecific attribute whose value in some instances can be used to predict its value in unknown instancesUnlabeled dataNo such specific attribute that can be used to predict the value in unknown instances. Supervised learning – Data mining using labeled dataUnsupervised learning – Data mining using unlabeled data
- 5. Labeled dataAttributes can be of two types:Categorical attribute Takes a value from only a fixed set of values (like an enumeration) eg. ‘very good’, ‘good’, ‘poor’Supervised learning is called ClassificationNumerical attributeCan take a value from a continuous range of numerical valuesSupervised learning is called Regression
- 6. Unlabeled dataIt doesn’t have any specifically designated attributeUnsupervised learning Data mining using unlabeled dataPurpose - To extract as much as it is possible from the data available.
- 7. Supervised learning: ClassificationIt is based on the following three methods:Nearest neighbor matching: Identifying the classified instances that are closest (in some sense) to the unclassified oneClassification rules:Look for rules that can be used to predict the classification of an unknown instanceClassification tree:Generation of classification rules via the tree-like structure
- 8. Supervised Learning: Numerical PredictionRegression is done by using Neural NetworksNeural Network: Given a set of inputs to predict one or more outputs
- 9. Unsupervised Learning: Association RulesAssociation Rules: To find any relationship that exists amongst the values of variables within a training setExample: IF variable_1>90 and switch_6 = openTHEN variable_3 < 47.5 and switch_9 = closed(probability = 0.8)
- 10. Unsupervised Learning: ClusteringTo find groups of items that are similarExample: A company may group its customers based on income to target its policies etc.
- 11. Visit more self help tutorialsPick a tutorial of your choice and browse through it at your own pace.The tutorials section is free, self-guiding and will not involve any additional support.Visit us at www.dataminingtools.net