数据分析方法课件

《数据分析方法课件》由会员上传分享，免费在线阅读，更多相关内容在教育资源-天天文库。

1、Chapter4TheMethodsofDataAnalysis6.1DatanormalizationDatanormalizationisthebasisforcomparingexperimentswithinlargeserieswhenexperimentalconditionsmaynotbeidentical.Normalizationensuresthattheexperimentalqualityofthedataiscomparableand,soundmathematicalalgo

2、rithmshavebeenemployed.Normalizationincludesvariousoptionstostandardizedataandtoadjustbackgroundlevelsandcorrectgradients.Thecommonlyusednormalizationfunctionsareasfollows:Linearnormalization:(6.1)Rationormalization:(6.2)Z-scorenormalization:(6.3)Generall

3、y,linearnormalizationisrecommended(ifX’max=1andX’min=0,x’iisnormalizedinpercentagebyformula(6.1)).AfterRationormalization,thesumofnormalizedvariableswillbeequalto1.Z-scoreassumesxiobeysGaussiandistribution.Ifxihasadifferentdistribution,thenthenormalizatio

4、nwilltwistthepattern(variancewillbefarawayfromthestandarddeviation)andleadstoincorrectpatternrecognition.σIspopulationstandarddeviation,ingeneral,itcanbeapproximatedbysamplestandarddeviation(S)6.2SimpleLinearRegressionLearningObjectives:1.DescribetheLinea

5、rRegressionModel2.ExplainOrdinaryLeastSquares3.ComputeRegressionCoefficients4.Evaluatethelinearregressionmodel5.PredictResponseVariable6.2.1DescribetheLinearRegressionModelRegressionModels:1.Answer‘WhatIstheRelationshipBetweentheVariables?’2.EquationUsed1

6、NumericalDependent(Response)Variable1orMoreNumericalorCategoricalIndependent(Explanatory)Variables3.UsedMainlyforPrediction&EstimationTypesofRegressionModelsRegressionModelsTypesofRegressionModelsRegressionModelsSimple1ExplanatoryVariableTypesofRegress

7、ionModelsRegressionModels2+ExplanatoryVariablesSimpleMultiple1ExplanatoryVariableTypesofRegressionModelsRegressionModelsLinearSimpleMultiple1ExplanatoryVariable2+ExplanatoryVariableTypesofRegressionModelsRegressionModelsLinearNon-LinearSimpleMultiple1Ex

8、planatoryVariable2+ExplanatoryVariableTypesofRegressionModelsRegressionModelsLinearNon-LinearSimpleMultipleLinear1ExplanatoryVariable2+ExplanatoryVariableTypesofRegressionModelsRegressionModelsLinearNon-LinearSimp