时间数列分析

TimeSeriesAnalysis2019年9月4日星期三O:/Network/Path/Filename.ppt-2-BenefitsandUsesofTimeSeries•Benefitsoftimeseries–Monitorsalesperformanceovertime…removevariationinmonthlysalescausedbycalendardifferencesandseasonalitythatcanconcealpotentialproblemswithsales–Accuratelydeterminethedirectionandrateofgrowth/declineinsales–Quicklyidentifychangesinsalestrendsandcorrelatethemtofactorsaffectingsales…industry,company,competition–Improvedecisionmakingregardingsalesandmarketingactions•Usesoftimeseries–Assesscurrentsalesperformanceandevaluatetheeffectivenessofsalesprograms–Determineunderlyingsalestrendandprojectyearendsales–EstablishappropriatebudgetsfornextyearandestimatemonthlybudgetspreadsTimeseriesanalysisistheprimarysalesanalysistechniqueatA-B2019年9月4日星期三O:/Network/Path/Filename.ppt-3-TimeSeriesAnalysis•WhatisTimeSeriesAnalysis?•HowareTimeSeriesplotsdeveloped?•WhataretheadvantagesofTimeSeriesAnalysis?•WhatareTimeSeriesusedfor?2019年9月4日星期三O:/Network/Path/Filename.ppt-4-WhatisTimeSeriesAnalysis?•Timeseriesanalysisisastatisticaltechniqueusedtoanalyzeandmonitorsalesvolumeovertime.2019年9月4日星期三O:/Network/Path/Filename.ppt-5-WhyTimeSeries?BeerSales050100150200250199819992000200120022003ThousandBarrels•Beersalesarehighlyseasonal•Itisverydifficulttoevaluatemonthlysalesovertime.2019年9月4日星期三O:/Network/Path/Filename.ppt-6-Howdotimeserieswork?•Monthlyvariationinsalesiscausedbytwomajorfactors–Seasonality–SellingDays(calendareffects)•TimeSeriestechniquestatisticallyremovestheeffectsofthesetwofactors•TimeSeriestechniqueusestheX-11procedureforseasonaladjustments–TheX-11procedurewasdevelopedbytheU.S.BureauofCensusinthe1950’s.ItwasbroughttoA-Bintheearly1960’sandhasbecomethestandardforreportingsales.2019年9月4日星期三O:/Network/Path/Filename.ppt-7-Howdotimeseriesadjustsales?•Asellingdayadjustmentfactorforeachmonthiscomputedandappliedtotherawsales–Thisfactorallowsyoutocomparemonthsasiftheyhadthesamenumberofsellingdays…e.g.accuratelycomparetheJunethisyearvs.Junelastyear•Aseasonalfactoriscomputedandappliedtothesellingdayadjustedsales–Thisfactor,whenapplied,givesyoumonthlydatadirectlycomparabletoanyothermonth…e.g.accuratelycompareJunethisyearwithMaythisyear2019年9月4日星期三O:/Network/Path/Filename.ppt-8-SellingDaysAllotherthingsbeingequal,salesinAug-03woulddecrease4.8%becauseofonelesssellingday.InordertocomparethetwomonthsAug-03saleswillhavetobeadjustedup+4.8%.SMTWTFSSMTWTFS121231.00.01.01.00.03456789456789100.01.01.01.01.01.00.00.01.01.01.01.01.00.010111213141516111213141516170.01.01.01.01.01.00.00.01.01.01.01.01.00.017181920212223181920212223240.01.01.01.01.01.00.00.01.01.01.01.01.00.024252627282930252627282930310.01.01.01.01.01.00.00.01.01.01.01.01.00.0310.0August2003August2002Aug-2003has21sellingdaysAug-2002has22sellingdays2019年9月4日星期三O:/Network/Path/Filename.ppt-9-Seasonality•Seasonalityisexpressedasanindexforamonthcomparedtoanaveragemonth.•Amonthwheresaleswere20%higherthanaveragewouldhaveaseasonalfactorof120.•Amonthwhichwas10%lowerthanaveragewouldhaveaseasonalfactorof90.JanFebMarAprMayJunNoSeasonality100100100100100100StrongSeasonality607580120140120JulAugSepOctNovDecNoSeasonality100100100100100100StrongSeasonality118807562120150020406080100120140160JanFebMarAprMayJunJulAugSepOctNovDecStrongSeasonalityNoSeasonality2019年9月4日星期三O:/Network/Path/Filename.ppt-10-AdjustingSalesRawSalesXSellingDayFactor÷SeasonalFactorSeasonallyAdjustedSales=MonthActualSales(Mbbls)SellingDayFactorSeasonalFactorAdjustedSalesJun-03211X1.004÷1.210=175Jul-03221X0.958÷1.212=175Aug-03196X1.054÷1.190=174Sep-03160X1.004÷0.948=1692019年9月4日星期三O:/Network/Path/Filename.ppt-11-Howdotimeserieswork?RawSalesSellingDayAdjustedSeasonallyAdjusted2019年9月4日星期三O:/Network/Path/Filename.ppt-12-DissectingaTimeSeriesPlot0200400600800100012001400160018002000199419951996199719981999AnnualizedSalesinMbblsAnnualizedSales…tellsushowbigthemarketis.TrendLine…tellsusthedirectionofsalesbasedonpast&presentperformanceIrregularvariations…showsustheimpactofmarketplaceactionsSTR’s;OntarioSTC’sDataDescription…tellsusthetypeofdataplotted2019年9月4日星期三O:/Network/Path/Filename.ppt-13-AdvantagesofTimeSeries•Advantagesoftimeseries:–Removesvariationinmonthlysalescausedbycalendardifferencesandseasonality–Helpustoaccuratelyestimatethedirectionandrateofsalesgrowth/decline–Theyareanimprovementoverothermethodssuchasyear-over-yeargrowthormovingaveragesbecausetheyshowuswhatishappeningsooner…anearlywarningofchangingsalesconditions•Timeseriessignificantlyimprovedecisionmaking…–Allowsustotakecorrectiveactionsooner–Allowsustotaketherightcorrectiveaction–Helpstoestablishappropriatesalesobjectives2019年9月4日星期三O:/Network/Path/Filename.ppt-14-AdvantagesofTimeSeries•Ifthetimeseriesshowsarelativesmoothpatternfromoneyeartothenext…thetrendandtheyearoveryeargrowthwouldprovideroughlythesamereading.•But,iftherewasasignificantmarketeventorchange,theyearoveryeartrendswillbemisleading051015202519981999+50%2019年9月4日星期三O:/Network/Path/Filename.ppt-15-MisleadingGrowthRates024681012141619981999PositiveTrend:Flat%Change0%024681012141619981999TrendFla