供应链需求预测的方法
for the forecast
時間序列預測
n Goal is to predict systematic component of demand
q Multiplicative: (level)(trend)(seasonal factor) q Additive: level + trend + seasonal factor q Mixed: (level + trend)(seasonal factor)
n Bottom up q 市調法Market research
n Long-range n New product sales q 歷史類推法Historical analogy n 類似的產品經驗類推 q Delphi Method n 以問卷方式蒐集專家意見以進行預測 n 經由問卷溝通,專家間無直接互動以避免主控性 n 以統計量收斂為停止指標
• Systematic component: Expected value of demand • Random component: The part of the forecast that deviates
from the systematic component • Forecast error: difference between forecast and actual demand
Systematic component (S) + Random component (R)
Level (current deseasonalized demand)
Trend (growth or decline in demand)
Seasonality (predictable seasonal fluctuation)
供应链需求预测的方法
2020年4月25日星期六
大綱
n 預測在供應鏈的角色 n 預測的特性 n 主要企業預測項目 n 預測的方法與組成 n 時間序列預測 n 預測誤差的衡量指標 n 執行預測的建議 n CPFR
預測在供應鏈的角色
n The basis for all strategic and planning decisions in a supply chain
需求資料組成的關係類型
n 相乘
q 系統部分=水準 ×趨勢 ×季節性因素
n 相加
q 系統部分=水準 + 趨勢 + 季節性因素
n 混合
q 系統部分=(水準 + 趨勢)× 季節性因素
時間序列預測
Forecast demand for the next four quarters.
時間序列預測
預測的方法
主要企業預測項目
n 市場需求量 n 母體數預測 n 單位需求量預測 n 驅動變數預測 n 市場佔有率預測 n 企業銷售量預測 n 單價預測 n 生命週期預測
預測的方法
n 主觀法(subjective methods)
預測人員依個人主觀的判斷進行預測 常應用在缺乏歷史資料時透過專家進行主觀預測 q 草根法Grass roots
n Static n Adaptive
q Moving average q Simple exponential smoothing q Holt’s model (with trend) q Winter’s model (with trend and seasonality)
預測的流程
n Understand the objectives of forecasting n Integrate demand planning and forecasting n Identify major factors that influence the
預測的方法
n 客觀法(objective methods)
以歷外插法)
n 假設過去之需求資料是未來需求良好指標下,使用歷史資料進 行預測,適合當需求環境穩定、無劇烈變動時進行
q Causal (因果關係法)
n 假設需求與環境中某些因素是高度相關,藉由發現需求與環境 因素的相關性去估計未來的需求
n Long-term forecasts are less accurate than short-term forecasts (forecast horizon is important)
n Aggregate forecasts are more accurate than disaggregate forecasts
planning q Personnel: workforce planning, hiring, layoffs
n All of these decisions are interrelated
預測的特性
n Forecasts are always wrong. Should include expected value and measure of error.
demand forecast n Understand and identify customer segments n Determine the appropriate forecasting
technique n Establish performance and error measures
q Transfer Function Model(轉換函數模式)
n 結合Time Series 與 Causal 兩者,經由解釋變數與應變數之 歷史資料產生轉換函數,再將解釋變數之預測值代入轉換函數 產生應變數之預測值
n ARIMAT 、SARIMAT
需求資料的組成
Observed demand (O) =
n Examples:
q Production: scheduling, inventory, aggregate planning q Marketing: sales force allocation, promotions, new
production introduction q Finance: plant/equipment investment, budgetary