# Forecasting Forecasting A Basis of Forecasting In business

Forecasting

預測 (Forecasting) • A Basis of Forecasting • In business, forecasts are the basis for budgeting and planning for capacity, sales, production and inventory, manpower, purchasing, and more. • Forecasting is not an exact science. • No single technique works all the time

預測 (Forecasting) • 對預測的認知 1. Forecasting techniques generally assume that the same underlying causal system that existed in the past will continue to exist in the future. 2. Forecasts are rarely perfect. 3. Forecasts for groups of items tend to be more accurate than forecasts for individual items. 4. Forecast accuracy decreases as the time period covered by the forecast - the time horizon increases.

預測 (Forecasting) • Forecasting 的步驟 1. Determine the purpose of the forecast 2. Establish a time horizon 3. Select a forecasting technique 4. Gather and analysis the appropriate data 5. Prepare the forecast 6. Monitor the forecast

預測 (Forecasting) • 預測方法 Forecasts Based on Judgement and Opinion (Qualitative Techniques) - No data is available • Forecasts Based on Time Series Data Associative Forecasts (Causal Forecasting) - Demand is related to some underlying factors in the environment

預測 (Forecasting) • Forecasts Based on Judgement and Opinion Executive opinions (marketing, product, engineering manufacturing, and finance) - Delphi method Consumer Surveys Opinions of the sales staff Opinions of experts Historical Analogy - Ties what is being forecast to a similar item

預測 (Forecasting) • Forecasts Based on Time Series Data Analysis of time series data requires the analyst to identify the underlying behavior of the series. 利用圖形作分析

預測 (Forecasting) • 時間序列的構成元件 Trend (趨勢性) -正向，負向，水平 S-Curve Trend Asymptotic Trend Exponential Trend Seasonality (季節性) Cycles (循環性) Irregular Variations (不正常變動) Random Variations (隨機性)

預測 (Forecasting) • 時間序列預測法 Naive Forecasts A naive forecast for any period equals the previous period’s actual value

預測 (Forecasting) • 時間序列預測法 Moving Average A moving average forecast uses a number of the most recent actual data values in generating a forecast. where n=number of periods in the moving average Ai =Actual value in period i MAt=Forecast of period t

Moving Average Actual MA 5 MA 3

預測 (Forecasting) • 時間序列預測法 Weighted Average is similar to a moving average, except that it assigns more weight to the most recent values in a time series.

預測 (Forecasting) • 時間序列預測法 Exponential Smoothing

Exponential Smoothing

預測 (Forecasting) • 時間序列預測法 Exponential Smoothing

預測 (Forecasting) • 時間序列預測法 Techniques of Trend • Linear Regression • Double Exponential Smoothing

預測 (Forecasting) • 時間序列預測法 Linear Regression

預測 (Forecasting) • 時間序列預測法 Linear Regression

預測 (Forecasting) • 時間序列預測法 Double Exponential Smoothing

預測 (Forecasting) • 時間序列預測法 Double Exponential Smoothing

預測 (Forecasting) • 時間序列預測法 Seasonal and Cyclic Demands When seasonal/cyclical effects are expected, the forecasting model can be developed to take them into effect using either multiplicative or additive factors

預測 (Forecasting) • 時間序列預測法 Multiplicative Model with Single Known Period Step 1: Separate the data into groups so that all members of a group belong to the same cycle. Step 2: Find the cycle average demand period average for each cycle. Step 3: Calculate the seasonal indices for each period in each cycle. This is done by the cycle average demand/ period average for that cycle.

預測 (Forecasting) • 時間序列預測法 Multiplicative Model with Single Known Period Step 4: Use the previous defined technologies to predict the cycle average demand/period for the next cycle. Step 5: Follow the reverse procedure to predict the demands in each period of this cycle.

預測 (Forecasting) • Associative Forecasts - Correlation Analysis Assume Y= a + b. X then Covariance=Cov(X, Y)=E(XY)-E(X)E(Y) If X and Y are independent, then cov(X, Y)=0. Correlation coefficient=

預測 (Forecasting) • Tracking Signal: Use to monitor the forecasting results Tracking Signal (TS) 一般而言 TS 值在 ± 4 之間為可容忍範圍

預測 (Forecasting) • Comparison of Forecasting Techniques

- Slides: 28