Lecture of forecasting short and updated 1.pptx

Forecasting lkllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll

1. Dr. Fauzi Zowid Supply chain and logistics management Lecture Demand Forecasting التنبؤ بالطلبيات Academic year of 2021-2022

2. Previous lecture Basics and principles of supply chain management. Phases of supply chain integration (idea, pros and cons). Importance of information availability, manufacturing processes based two strategies (make to stock or make to order). Discussed (Decisions levels, horizon times and examples) as an approach to build the course contents as follows. Demand forecasting system S&OP system MRP and BOM systems MPS system ERP system Inventory and stock control system Distribution strategies and system Transportation, customer value and smart pricing 2

3. Agenda Demand forecasting (definition, objectives and importance). Approaches and methods to forecasting Moving Average Weighted Moving Average Exponential Smoothing Linear Regression. Forecasting accuracy/validation. Summary 3

4. Forecasting What is Forecasting Forecasts are a critical part of the operation manager’s function and decision–making. (importance). Understanding forecasting time horizons. Forecasting patterns. Quantitative vs qualitative forecasting techniques. How mathematically calculate forecasts? Knowing software such as Excel, SAP, SPSS… etc are as usage options. Learning Objectives 4

5. Forecasting usage and examples In production and operations department, marketing department, human resource department, finance department and accounting department E.g. weather, earthquake, stock prices, number of sales, number of productions and inventories. Usage Examples 5

6. Basics and Principles of Forecasting Forecasting Methods Forecast Errors and Accuracy Forecasting

7. Forecasting 7 هي عملية تقدير سلوك المبيعات او الطلبات المستقبلية اعتماداً على سلوكها في الفترة الماضية. A technique that is used for the estimation of what can be the demand for the upcoming product or services in the future.

8. Forecasting Forecasting can be defined as “ the art and science of predicting future events or trends for planning purposes ”. Business Forecasting: used to predict profits, revenues, costs, productivity changes, prices & availability of energy & raw materials, interest rates, movements of key economic indicators, & prices of stocks and bonds. Our Focus: Forecasting Demand No single technique works all the time 8

9. Why demand Forecasting? Demand Forecasts are Major inputs for many Operations Decisions Supports the decisions on facilities/capacity planning and budget. Supports the decisions on aggregate production planning and MRP. Supports other organizational functions (HR, information, funds or financing). Helps to evaluate the market (opportunities and threads) for future investments. 9

10. Elements of good forecasting Timeliness Forecasting horizon must cover the time needed to make potential changes Accuracy Degree of Accuracy (or Confidence Level) Reliability (Robustness) Forecasting Technique(s) must consistently provide good results Meaningful (Units) Financial Planners  Dollars Production Planners  units of product Schedulers  machines, skills, hours Simplicity 10

11. Steps in Forecasting Academic Year of 2019-2020 11 Determine the Purpose of forecast Why ? When ? the level of details? degree of accuracy? allocation of resources? Determine the Time Horizon Accuracy decreases as time horizon increases Select the Forecasting Technique(s) Accuracy required? Time horizon? Resources / data availability? Gather & Analyze Proper Data Prepare Forecast Monitor Forecast Make sure forecast is working satisfactorily No change in underlying conditions (e.g. market conditions, political environment, customer preferences)

12. Types of Demand A B(4) C(2) D(2) E(1) D(3) F(2) Dependent Demand : Raw Materials, Component parts, Sub-assemblies, etc. Independent Demand : Finished Goods 12 Figure 1: an illustrative draw of dependent and independent demands

13. Forecasting Time Frames & Patterns Forecasting Time Frames are as: Short-range (one month to less than one year) Medium-range (above one year but less five years) Long-range (more than five years) ** Short-Range Forecasts are more accurate than Long-Range Forecasts Forecasting Patterns: Horizontal. Trend ( e.g : stock prices) Random variations (cannot be predicted) Cycles ( e.g : automobile sales, smart devices sales) Seasonal pattern ( e.g : every spring demand or every summer demands) 13

14. Patterns of Demand Quantity Time (a) Horizontal : Data cluster about a horizontal line. Quantity Time (b) Trend : Data consistently increase or decrease. Quantity Months from Jan to Dec. Year 1 Year 2 (c) Seasonal : Data consistently show peaks and valleys. Quantity | | | | | | Years (c) Cyclical : Data reveal gradual increases and decreases over extended periods. 14 Figure 2: various patterns of demand

15. Forecasting Forecasting methods Forecast Errors and Accuracy Forecasting

16. Forecasting Methods Three basic types of forecasting methods: Times Series - Statistical techniques that use historical data to predict future behavior . Regression Method - attempt to develop a mathematical relationship between the item being forecast (dependent variables) and factors (independent variables) that are assumed to cause the results observed in the past. Qualitative Methods - Methods using judgment, expertise and opinion, historical analogy or customer’s opinion to make forecasts decisions. 16

17. Time Series Method Statistical techniques that use of historical data collected over a long period of time. Methods assume that what has occurred in the past will continue to occur in the future. Forecasts based on only one factor - time . 17

18. Time Series Method Classifications of Time Series Methodology: Moving Average Method Naïve Method Weighted Moving Average. Exponential Smoothing. Adjusted Exponential Smoothing. Winter’s model Seasonal Adjustments. “Seasonality” Linear regression/trend 18

19. Time Series Method (Moving Average) Moving Average is “a forecasting method that uses an average of demands in a recent period to forecast the next period”. Formula of moving average method Where is demand in period t divided by n as number of periods in the moving average Examples Three-period moving average Five-period moving average   19

20. Time Series Method (Moving Average) Figure 3 Three- and Five-Month Moving Averages 20

21. Time Series Method (Weighted Moving Average) In weighted moving average, weights are assigned to the demands in most recent periods to forecast the next period. formulas of weighted moving average method are as:   Weights are equal to 1 or can be assigned based on experience and intuition. Examples: Three- period weighted moving average Five- period weighted moving average *** Mean Absolute Deviation (MAD) technique can be used in order to make a comparison of forecasting methods and measure errors.   21

22. Time Series Method Advantages : Ease of computation Low computational cost Ease of understanding Fast to generate forecast results Disadvantages : Slow Response to changes It fails to produce accurate forecasts if the data follows seasonal, random and cyclical patterns. Requires carrying ‘large’ amount of data. 22

23. Naïve Method Naïve method: A forecasting technique which assumes that demand in the next period is equal to demand in the most recent period. Formula of naïve method: where, is equal to a data period in most recent period t. Advantages : No Cost Quick & Easy to Prepare & Understand Disadvantages : Cannot provide highly accurate forecast Traces actual data (no smoothing)   23

24. Exponential smoothing method Exponential smoothing is a sophisticated weighted moving average that calculates the average of a time series by giving recent demands more weight than earlier demands. Requires only three items of data The last period’s forecast for this period The demand for this period A smoothing parameter, alpha ( α ), where 0 ≤ α ≤ 1.0 The equation for the forecast is Or Or   24 F t +1 = F t + α ( D t – F t )

25. Exponential smoothing method 25 Figure 4 Exponential smoothing forecasts

26. Time series methods 26 Naïve equation: Moving Average: Weighted Moving Average: Exponential smoothing :   F t +1 = F t + α ( A t – F t )

27. Linear regression method Linear regression relates demand (dependent variable ) to an independent variable . When demand displays an obvious trend over time, a least squares regression line , or linear trend line , can be used to forecast. Formula: 27

28. Linear regression method 28

29. Forecast Errors and Accuracy Forecast Error = Actual Demand – Forecast Demand E t = A t – F t E t > 0  Forecast lower than actual E t < 0  Forecast higher than actual Other measures of forecast errors: Mean Absolute deviation (MAD) Mean absolute percentage deviation (MAPD) Mean Absolute Percent Error (MAPE) Mean Squared Error (MSE) Running Sum of Forecast Errors (RSFE) Standard deviation (  ) Forecasts will always deviate from actual values. Would like forecast error to be as small as possible. 29

30. Measures of Forecast Errors E t = A t - F t  | E t | n  E t 2 n RSFE =  E t  = MSE = MAD = MAPE =  [ | E t | / A t ] * 100 n  E t 2 n 30 MAPD =  | D t - E t | /  D t

31. Issues in demand forecasting methods 31 What type of a time series method to implement? How many periods to use? How to determine/assign periods weights? How to measure the validity of a forecasting system?

32. Example 1 If the actual number of demands in month 5 is 805, what is the forecast for month 6 using Naïve 3-period moving average methods simple estimate of forecasting errors? 32 Month Demand 1 800 2 740 3 810 4 790 5 805

33. Example 1 If the actual number of demands in month 5 is 805, what is the forecast for month 6 using: Naïve and simple estimate of forecasting errors? 33 Month Demand 1 800 2 740 3 810 4 790 5 805 F 6 errors = 805 - 805 = 0 F 6 = 805

34. Example 1 If the actual number of demands in month 5 is 805, what is the forecast for month 6 using: 3-period moving average methods and simple estimate of forecasting errors? 34 Month Demand 1 800 2 740 3 810 4 790 5 805 F 6 = = 802 D 5 + D 4 + D 3 3 805 + 790 + 810 3 = F 6 errors = 805 - 802 = 3

35. Example 1 Suppose that there were 790 demands in month 4 ( A t ), whereas the forecast ( F t ) was for 783 demands. Use exponential smoothing with α = 0.20 to compute the forecast for month 5. Plus, estimate the forecast errors. 35 Month Demand 1 800 2 740 3 810 4 790 5 805 F t +1 = F t + α ( A t – F t ) 783 + 0.20(790 – 783) = 784 F 5 errors = 805 – 784 = 21 Given the number of actual demand 805 ,

36. Problem 2 (Comparison) Last year’s sales of a product at a retail store are shown in the following table: Month Actual Sales (products) January 200 February 250 March 180 April 190 May 175 June 170 July 180 August 210 September 220 October 200 November 200 December 205 36

37. Problem 2 (Comparison) 3-period moving average method and 3-period weighted moving average method (Weight 1 month ago = 5, Weight 2 months ago = 3, and Weight 3 months ago = 1) 37 Month Actual Sales Forecast (a) Absolute Deviation (a) Forecast (b) Absolute Deviation (b) Jan. 200         Feb. 250         Mar. 180         April 190 180+250+200/3 =210 210-190= 20 (180*5)+(250*3)+(200*1)/5+3+1=206 206-190= 16 May 175 190+180+250/3=207 207-175= 32 (190*5)+(180*3)+(250*1)/5+3+1=193 18 Jun. 170 182 12 181 11 Jul. 180 178 2 174 6 Aug. 210 175 35 176 34 Sept. 220 186 34 196 24 Oct. 200 203 3 212 12 Nov. 200 210 10 208 8 Dec. 205 207 2 202 3       Mean = 17   Mean = 15

38. Problem 3 The Polish General’s Pizza Parlor is a small restaurant catering to patrons with a taste for European pizza. One of its specialties is Polish Prize pizza. The manager must forecast weekly demand for these special pizzas so that he can order pizza shells weekly. Recently, demand has been as follows: Week الفترة Pizzas demand الطلبيات للبيتزا 1 50 2 65 3 52 4 56 5 55 6 60 38

39. Problem 3 Solutions 39 Week Pizzas Simple Moving Average (Forecast) Absolute Deviation Weighted Moving Average (Forecast) Absolute Deviation 1 50         2 65         3 52         4 56 52+65+50/3= 56 56-56= 0 (52*.50)+(65*.30)+(50*.20)/.50+.30+.20=56 56-56= 0 5 55 56+52+65/3= 58 3 (56*.50)+(52*.30)+(65*.20)/.50+.30+.20=57 2 6 60 55+56+52/3= 54 6 55 5       Mean = 3   Mean = 2

40. Example 5 Computer services demand and periods are given in the following table: 40 Compute the correlation between the two variables.

41. Example 5 41 Solution:

42. Forecasting Forecasting Methods Forecast Errors and Accuracy Forecasting

43. Be familiar with… Forecast supports operations and supply chain decisions: How much to produce… How much to stock… How much to order… Forecasting time horizons categories: Short-term forecasts Mid-term forecasts Long-term forecasts In forecasting, no specific technique will result in accurate forecast. Short-term forecasting is more suitable and accurate 43

44. Be familiar with… Managers can predict for future by using judgment, opinion, or past experience . “ Qualitative method” Supporting techniques for forecasting include the Delphi method, market research, customer surveys, etc. Managers can predict for future by using various mathematical methods for decision making. We Do forecast for Human resources (hiring or training, … etc ) Marketing (pricing, new promotions/products, … etc.) Supply chain (capacity of demand, dealing with suppliers) Finance (profits, revenues, cash flows,… etc.) 44

45. On Academic Researches 45 Demand and intermittent demand forecasting… (hot topic for academic and real practices in enterprises) General research directions: Demand forecasting performance. Investigating or Developing an optimization method/s to forecasting intermittent demand. (Service level, costs and errors) Considering either slow-fast moving or Intermittent demand forecasts link with inventory classification and stock control. (Service level, costs and errors) Impact of intermittent demand forecasts on inventory control polices Empirical investigation of Demand forecasting performance during the Covid-19 pandemic.

46. Class participation 46 To get (2.5 marks) Solve various demand forecasting problems based on: Exponential smoothing Moving average method Linear regression. Send your solutions in a (Word file) to the email address: [email protected] Name email subject as: “SCM: Demand Forecasting” Submission date: one week after discussing the lecture. All the best…!

47. Take-away Demand Forecasting for Operations and Supply Chain planning. Notes and examples Read chapter 7 Lecture Your job ***Keep enjoying the course..!*** 47