數(shù)據(jù)挖掘概念與技術(shù)CHAPTER2-了解數(shù)據(jù)

《數(shù)據(jù)挖掘概念與技術(shù)CHAPTER2-了解數(shù)據(jù)》由會(huì)員上傳分享，免費(fèi)在線閱讀，更多相關(guān)內(nèi)容在教育資源-天天文庫。

1、DataMining:ConceptsandTechniques楊昆修譯—Chapter2—JiaweiHan,MichelineKamber,andJianPeiUniversityofIllinoisatUrbana-ChampaignSimonFraserUniversity1Chapter2:了解數(shù)據(jù)數(shù)據(jù)對象和屬性類型DataObjectsandAttributeTypes數(shù)據(jù)的(基本)統(tǒng)計(jì)描述BasicStatisticalDescriptionsofData數(shù)據(jù)可視化DataVisualization測量數(shù)據(jù)相似性和相異性MeasuringDataSimilari

2、tyandDissimilarity總結(jié)Summary2數(shù)據(jù)集合的類型記錄Record關(guān)系記錄數(shù)據(jù)矩陣,e.g.,數(shù)值矩陣,交叉表文檔數(shù)據(jù):文本文檔:詞頻向量term-frequencyvector交易數(shù)據(jù)圖and網(wǎng)絡(luò)萬維網(wǎng)社會(huì)或信息網(wǎng)絡(luò)分子結(jié)構(gòu)MolecularStructures有序的Ordered視頻數(shù)據(jù):sequenceofimages時(shí)間數(shù)據(jù):時(shí)間序列time-series序列數(shù)據(jù):交易序列transactionsequences遺傳序列數(shù)據(jù)空間,圖像imageand多媒體multimedia:Spatialdata:mapsImagedata:Videodata:

3、3結(jié)構(gòu)數(shù)據(jù)的重要特征維度Dimensionality維數(shù)災(zāi)難Curseofdimensionality稀疏Sparsity只有計(jì)數(shù)Onlypresencecounts分辯率Resolution模式依賴于尺度分布Distribution中心性和分散Centralityanddispersion4數(shù)據(jù)對象數(shù)據(jù)集由數(shù)據(jù)對象構(gòu)成一個(gè)數(shù)據(jù)對象代表一個(gè)實(shí)體例子:銷售數(shù)據(jù)庫salesdatabase:客戶/顧客,商店物品,sales醫(yī)學(xué)數(shù)據(jù)庫:patients,treatments大學(xué)數(shù)據(jù)庫:students,professors,courses又稱為樣本,事例,實(shí)例,數(shù)據(jù)點(diǎn),對象,元組t

4、uples.數(shù)據(jù)對象由屬性來描述Databaserows->dataobjects;columns->attributes.5屬性屬性Attribute(or維度,特征,變量):一個(gè)數(shù)據(jù)字段,表示一個(gè)數(shù)據(jù)對象的某個(gè)特征.E.g.,customer_ID,name,address類型:名詞性Nominal二元的數(shù)字的Numeric:數(shù)量的Interval-scaledRatio-scaled6屬性類型名詞性Nominal:類別,狀態(tài),or“名目”Hair_color={auburn,black,blond,brown,grey,red,white}婚姻狀態(tài),職業(yè)occupati

5、on,IDnumbers,zipcodes二元只有2個(gè)狀態(tài)的名詞性屬性(0and1)對稱二元Symmetricbinary:同樣重要的兩相e.g.,gender非對稱Asymmetricbinary:非同等重要e.g.,醫(yī)療檢查(positivevs.negative)慣例Convention:assign1tomostimportantoutcome(e.g.,HIVpositive)順序的Ordinal值有一個(gè)有意義的順序(排序)但連續(xù)值之間的大小未知.Size={small,medium,large},等級(jí),軍隊(duì)排名7數(shù)值屬性的類型數(shù)量Quantity(integero

6、rreal-valued)區(qū)間Interval在某個(gè)同等大小的一個(gè)尺度單位上Measuredonascaleofequal-sizedunits值有序E.g.,temperatureinC?orF?,calendardates沒有真正的零點(diǎn)Ratio有真正的零點(diǎn)可以講值是被測量單位一個(gè)數(shù)量級(jí)(10K?istwiceashighas5K?).e.g.,溫度在開爾文,長度,計(jì)數(shù),貨幣的數(shù)量8離散vs.連續(xù)屬性DiscreteAttribute一個(gè)有限的或可數(shù)無限集值E.g.,zipcodes，thesetofwordsinacollectionofdocuments有時(shí),表示為整

7、數(shù)變量注:二元屬性是離散屬性的一個(gè)特殊情況ContinuousAttribute屬性值為實(shí)數(shù)E.g.,temperature,height,orweight實(shí)際上，實(shí)值只能使用有限位數(shù)進(jìn)行測量和代表連續(xù)屬性通常表示為浮點(diǎn)變量9Chapter2:數(shù)據(jù)的統(tǒng)計(jì)描述DataObjectsandAttributeTypes數(shù)據(jù)的(基本)統(tǒng)計(jì)描述數(shù)據(jù)可視化測量數(shù)據(jù)相似性和相異性MeasuringDataSimilarityandDissimilaritySummary10數(shù)據(jù)的(基本)統(tǒng)計(jì)描述Motivation為