In theretailstageofafoodsupplychain

In theretailstageofafoodsupplychain,foodwasteandstock-outsoccurmainlyduetoinaccurate
forecasting ofsaleswhichleadstoincorrectorderingofproducts.Thetimeseriessalesinfoodretail
industry arecharacterizedbyhighvolatilityandskewness,whichvarybytime.So,theintervalforecasts
are requiredbytheretailcompaniestosetappropriateinventorypolicy(reorderpointorsafetystock
level). Thispaperattemptstodevelopaseasonalautoregressiveintegratedmovingaveragewithexternal
variables(SARIMAX)modeltoforecastdailysalesofaperishablefood.Theprocessof fitting aSARIMAX
model inthisstudyinvolves:(i)thedevelopmentofSeasonalAutoregressiveIntegratedMovingAverage
(SARIMA) modeland(ii)combiningtheSARIMAmodelandthedemandinfluencing factorsusinglinear
regression.AstheSARIMAXusingmultiplelinearregression(SARIMA-MLR)modelproducesonlymean
forecast, thepossibilityofunderestimationandoverestimationisveryhighduetohighservicelevel,
peak,andsparsesalesinfoodretailindustry.Therefore,ahybridSARIMAandQuantileRegression
(SARIMA-QR) isdevelopedtoconstructhighandlowquantilepredictions.Insteadofextrapolatingthe
quantilesfromthemeanpointforecastsofSARIMA-MLRmodelbasedontheassumptionofnormality,
the SARIMA-QRmodeldirectlyforecaststhequantiles.ThedevelopedSARIMA-MLRandSARIMA-QR
models areappliedinmodelingandforecastingofsalesdata,i.e.,thedailysalesofbananafromadis-
count retailstoreinLowerBavaria,Germany.TheresultsshowthattheSARIMA-MLRand-QRmodels
yield betterforecastsatout-sampledatawhencomparedtoseasonalnaïveforecasting,traditional
SARIMA, andmulti-layeredperceptronneuralnetwork(MLPNN)models.UnliketheSARIMA-MLRmodel,
the SARIMA-QRmodelprovidesbetterpredictionintervalsandadeepinsightintotheeffectsofdemand
influencing factorsfordifferentquantiles.

ผลลัพธ์ (ไทย) 1:

ใน theretailstageofafoodsupplychain, foodwasteandstock-outsoccurmainlyduetoinaccurateการคาดการณ์ ofsaleswhichleadstoincorrectorderingofproducts Thetimeseriessalesinfoodretailอุตสาหกรรม arecharacterizedbyhighvolatilityandskewness, whichvarybytime ดังนั้น theintervalforecastsมี requiredbytheretailcompaniestosetappropriateinventorypolicy (reorderpointorsafetystockระดับ) Thispaperattemptstodevelopaseasonalautoregressiveintegratedmovingaveragewithexternalตัวแปร (SARIMAX) modeltoforecastdailysalesofaperishablefood Theprocessof aSARIMAX ที่เหมาะสมรุ่น inthisstudyinvolves: (i) thedevelopmentofSeasonalAutoregressiveIntegratedMovingAverage(สมถวิล) modeland (ii) combiningtheSARIMAmodelandthedemandinfluencing factorsusinglinearถดถอย AstheSARIMAXusingmultiplelinearregression(SARIMA-MLR) modelproducesonlymeanการพยากรณ์ thepossibilityofunderestimationandoverestimationisveryhighduetohighservicelevelสูงสุด andsparsesalesinfoodretailindustry ดังนั้น ahybridSARIMAandQuantileRegressionIsdevelopedtoconstructhighandlowquantilepredictions (สมถวิล-QR) InsteadofextrapolatingthequantilesfromthemeanpointforecastsofSARIMA-MLRmodelbasedontheassumptionofnormalityสมถวิล-QRmodeldirectlyforecaststhequantiles.ThedevelopedSARIMA-MLRandSARIMA-QRรุ่น areappliedinmodelingandforecastingofsalesdata,i.e.,thedailysalesofbananafromadis-จำนวน retailstoreinLowerBavaria,Germany.TheresultsshowthattheSARIMA-MLRand-QRmodelsผลผลิต betterforecastsatout-sampledatawhencomparedtoseasonalnaïveforecasting ดั้งเดิมสมถวิล รุ่น andmulti layeredperceptronneuralnetwork (MLPNN) UnliketheSARIMA-MLRmodelสมถวิล-QRmodelprovidesbetterpredictionintervalsandadeepinsightintotheeffectsofdemandมีอิทธิพลต่อ factorsfordifferentquantiles

ผลลัพธ์ (ไทย) 2:

ใน theretailstageofafoodsupplychain, foodwasteandstock-outsoccurmainlyduetoinaccurate
คาดการณ์ ofsaleswhichleadstoincorrectorderingofproducts.Thetimeseriessalesinfoodretail
arecharacterizedbyhighvolatilityandskewness อุตสาหกรรม whichvarybytime.So, theintervalforecasts
มี requiredbytheretailcompaniestosetappropriateinventorypolicy (reorderpointorsafetystock
ระดับ) Thispaperattemptstodevelopaseasonalautoregressiveintegratedmovingaveragewithexternal
ตัวแปร (SARIMAX) modeltoforecastdailysalesofaperishablefood.Theprocessof กระชับ aSARIMAX
inthisstudyinvolves รุ่น: (i) thedevelopmentofSeasonalAutoregressiveIntegratedMovingAverage
(SARIMA) modeland (ii) combiningtheSARIMAmodelandthedemandinfluencing factorsusinglinear
regression.AstheSARIMAXusingmultiplelinearregression (SARIMA-MLR) modelproducesonlymean
คาดการณ์ thepossibilityofunderestimationandoverestimationisveryhighduetohighservicelevel,
ยอด andsparsesalesinfoodretailindustry.Therefore, ahybridSARIMAandQuantileRegression
(SARIMA QR- ) isdevelopedtoconstructhighandlowquantilepredictions.Insteadofextrapolatingthe
quantilesfromthemeanpointforecastsofSARIMA-MLRmodelbasedontheassumptionofnormality,
SARIMA-QRmodeldirectlyforecaststhequantiles.ThedevelopedSARIMA-MLRandSARIMA QR-
รุ่น areappliedinmodelingandforecastingofsalesdata คือ thedailysalesofbananafromadis-
นับ retailstoreinLowerBavaria, Germany.TheresultsshowthattheSARIMA-MLRand-QRmodels
ผลผลิต betterforecastsatout-sampledatawhencomparedtoseasonalnaïveforecastingแบบดั้งเดิม
SARIMA, andmulti-layeredperceptronneuralnetwork (MLPNN) รุ่น .UnliketheSARIMA-MLRmodel,
SARIMA-QRmodelprovidesbetterpredictionintervalsandadeepinsightintotheeffectsofdemand
ที่มีอิทธิพลต่อ factorsfordifferentquantiles