END 101 Service Systems and Forecasting Models PDF

1 END 101 Service Systems and Forecasting Models Assoc. Prof. Eda YÜCEL TOBB ETÜ  2 Service Systems  3 Operations Management History Operations Management: Set of activities that c...

1 END 101 Service Systems and Forecasting Models Assoc. Prof. Eda YÜCEL TOBB ETÜ  2 Service Systems  3 Operations Management History Operations Management: Set of activities that create value in the form of goods and services Timeline of OM: Crafts Manufacturing Industrial Revolution (1770 England): Invention of steam engines, replaced human power Started with textile industry, Increased use of refined coal Division of Labor (1776, Adam Smith) Standardized parts (1800, Eli Whitney), interchangable parts Steam Ships (1807) Telegraph (1844, Samuel Morse) Telephone (1876, Graham Bell) Scientific Management (1881, Frederick Taylor) Coordinated Assembly Line (1913, Henry Ford) First used moving assembly line Mass production in auto industry  4 Operations Management History Timeline of OM continued: Motion Study (1922, Frankard Gillian) Inventory Management (1915, F.W. Harris) Quality Control (1930s, Roming, Shewart, Dodge) Operations Research Applications in Warfare (1940s) Commercial Digital Computers (1951, IBM) Extensive Development of OR Tools in business: Simulation, Waiting Line Theory, Material Requirement Planning (MRP) Service Quality and Productivity (1970s) Emphasis on Quality (1980s-1990s) Just in time, Lean manufacturing Home and Personel Computers (1981, IBM) Internet (1990s) Supply Chain Management (1990s) ERP, SAP, Oracle E-Commerce and Service Science (2000s)  5 Introduction We are witnessing the greatest labor migration since the industury evolution From agriculture and manufacturing to services Migration is driven by global communications, business and technology growth, urbanization, and low cost labor  6 Introduction Service industries are leaders in every industrialized nation They create new jobs that dominate national economies Have the potential to enhance the quality of life of everyone  7 Service Sector And Economy One important reason for the interest in service sector is the quality trend that started in 1980s resulting in increase in demand for quality. Service sector continuously increased its share in GDP and job market.  8 Definitions A service is a time-perishable, intangible experience performed for a customer acting in the role of a co-producer. James Fitzsimmons Service enterprises are organizations that facilitate the production and distribution of goods, support other firms in meeting their goals, and add value to our personal lives. James Fitzsimmons  9 Products and Services Almost all product purchases comes with services A TV is a product, but is it possible to use TV without broadcasting service?  10 Products and Services Consider a custom made car producer that works with the customer from beginning to the end. Then what type of a firm would that be, a manufacturing firm or a service firm?  11 Products and Services The difference between service and products is not clear most of the time. In fact, most services are a mixture of product and service. Similarly, most products comes with service. If there is no product involved in a service, it is called “pure service”.  12 Product And Services Renting a car requires service but car itself is a product While food consumed at a restaurant is a product, having a nice dinner is a part of service. Every purchase is usually a package consisting of a mixture of service and products. Car manufacturers realized, it is more profitable to maintain the car not just sell it. Similarly, elevator manufaturers earn more from the maintanance service  13 Role of Services in an Economy FINANCIAL SERVICES INFRASTRUCTURE SERVICE · Financing · Communications · Leasing · Transportation · Insurance · Utilities · Banking PERSONAL SERVICES MANUFACTURING · Healthcare Services inside company: · Restaurants · Finance DISTRIBUTION · Hotels · Accounting SERVICES · Legal · Wholesaling · R&D and design · Retailing · Repairing CONSUMER (Self-service) BUSINESS SERVICES · Consulting GOVERNMENT SERVICES · Auditing · Military · Advertising · Education · Waste disposal · Judicial · Police and fire protection Services are not peripheral activities but rather integral parts of the society!  14 Economic Evolution During the past 90 years,we have witnessed a major evolution in our society from being predominantly manufacturing-based to being predominantly service- based. In the early 90s, 3 of 10 workers in US were employed in services sector, but today, services employ about 8 of 10 workers. The following figure shows the rapid increase in service employment in the US over the past century.  15 Trends in U.S. Employment by Sector Services: Value from enhancing the capabilities and interactions among people Goods: Percent Value from making a product Agriculture: Value from harvesting nature Year 2012 1-15 Percent Distribution of GDP List of Countries by GDP Sector Composition for year 2017 Kaynak: http://statisticstimes.com/economy/co untries-by-gdp-sector- composition.php Bir ekonominin tamamı ile hizmet sektörüne dayalı olması mümkün müdür? Bir ekonominin tamamı ile hizmet sektörüne dayalı olması mümkün müdür?  19 Stages of Economic Development Sociologist Daniel Bell (1973): Categorizes the development of societies: Preindustrial Society: Game against nature - Labor force is angaged in agriculture, mining and fishing - They are agrarian and structured arround tradition, routine and authority Industrial Society: Making more with less - The predominant activity is the production of goods - A world of cities, factories and tenements - Division of labor is the major law - The rhythm of life is a machine-paced and dominated by rigid working hours and time clocks Postindustrial Society: Concern is quality of life - Quality of life as measured by services such as health, education and recreation - Consumption and production occur simultaneously for services - Demand for them is more stable than that for manufactured goods  20 New Experience Economy Pine and Gilmore (1998): Experience Economy A stage of economic evolution in which added value is created by engaging and connecting with the customer in a personal and memorable way. The nature of service economy has moved past the transactional nature of services to one of experience-based relationship. ¢ Agrarian Economy : Harvesting coffee ¢ Industrial Economy: Packing and selling coffee ¢ Service Economy: Serving coffee in a shop ¢ New Experience Economy: Selling coffee at Starbucks o Companies define their respective services as experience.  21 New Experience Economy The Experience economy is further divided into “consumer services” and “business services” Consumer Service Experience Experiences create added value by engaging and connecting with the customer: Movie, Scuba Diving, Tourism Bussiness Service Experience Business to Business Services (B2B) as in a consultancy. Co-creation of value: Customer is a coproducer of the value Customer is an input to the service process Quality of service is measured primarily from the perspective of the customer  22 Source of Service Sector Growth Service sector is fueled by advances an information technology, innovation, and changing demographics Information Technology - Removes the need for physical proximity for service delivery (Online Banking) - Impacts the process of service delivery and created new service value chains with new business opportunities (Online Shopping) - In the future, major part of USA GDP will be generated by “information chains” not supply chains.  23 Source of Service Sector Growth Innovation Push theory (driven by technology and engineering, e.g. Post-it invented at a laboratory) Pull theory (e.g. Cash Management, airport shuttle service) Information driven services Product Innovations: Lead to the creation of new industries Process Innovations: Lead to increased efficiency Social Trends (Changing Demographics) Aging of the population (Hospitals) Two-income families (daycare) Growth in number of single people Home as sanctuary  24 Distinctive Characteristics of Service Operations 1. Customer Participation in the Service Process 2. Intangibility 3. Simultaneity 4. Perishability and Variability 5. Heterogeneity 6. Nontransferrable Ownership  25 Distinctive Characteristics of Service Operations 1. Customer Participation in the Service Process Attention to facility design e.g. Automobile industry Customer having an active role e.g. Health, Education On the other hand, taking customers out of the ‘physical’ process is becoming a common practice e.g. Retail Banking, ATMs  26 Distinctive Characteristics of Service Operations 2. Intangibility Services are ideas and concepts; Products are things, To secure a novel service, the firm must expand rapidly Franchising: to secure market areas A problem: Different from a product purchase, for the performance of the service, customer must rely on the reputation of the service firm Through use of registration, licensing and regulations, governments assure customers that some service providers assure certain standards  27 Distinctive Characteristics of Service Operations 3. Simultaneity Services cannot be stored Factory: a closed system; can be operated at a constant output rate Services: open systems; demand fluctuation Inventory control in manufacturing systems Customer waiting and queuing in service systems Service capacity Facility utilization Use of idle time A regular quality control (as in manufacturing) cannot be applicable.  28 Distinctive Characteristics of Service Operations 4. Perishability and Variability Perishable commodity, cannot be stored e.g. an empty airline seat, unoccupied hospital or hotel room Variable demand Consumer demand for services typically exhibit cyclic behavior over short period of time (e.g. daily, monthly) Restaurant lunch services between 12 am – 1 pm Demand for call centers Managers’ approaches to manage perishability and variability Smooth demand by using reservations, price incentives, etc. Adjust service capacity by using part-time, overtime, customer self- service Allow customers to wait  29 Distinctive Characteristics of Service Operations 5. Heterogeneity Service delivery may vary from customer to customer Customers expect to be served fairly Standards, employee trainings, feedback systems, happy employers “In the service business, you can’t make happy guests with unhappy employers.” J.Willard Marriot  30 Distinctive Characteristics of Service Operations 6. Nontransferrable ownership Unlike goods, services do not involve transfer of ownership Customers do not purchase an asset but use an asset for a specific time e.g. Hotel room for a night, car rental, house rental, seat on an airplane, sports center membership Sharing resources among customers presents management challenges  31 The Service Package Supporting facility Implicit Service Service Facililating Explicit Services goods Information  32 The Service Package Ha 1. Supporting Facility: The physical resources that must st an be in place before a service can be sold. Examples are e golf course, ski lift, hospital, airplane. 2. Facilitating Goods: The material used during the service. Might be purchased or consumed by the buyer or provided by the service producer. Examples are food items, legal documents, golf clubs, medical supplies, car parts. 3. Information: Operations data or information that is provided by the customer to enable efficient and customized service. Examples are patient medical records, customer preferences, location of customer to dispatch a taxi.  33 The Service Package 4. Explicit Services: Benefits readily observable by the senses. The essential or intrinsic features. Examples are quality of meal, on-time departure of train/bus, response time of a fire department, absence of pain after a tooth repair, a smooth- running car after a tuneup 5. Implicit Services: Psychological benefits or extrinsic features which the consumer may sense only vaguely. Examples are security of a well lighted parking lot, privacy of loan officer, atmosphere of a restaurant Competitive Service Strategies 34 Competitive Service Strategies Three general strategies have been successful in providing a competitive advantage: (1) Overall Cost Leadership (2) Differentiation (3) Focus 3-34 Competitive Service Strategies 35 Competitive Service Strategies 1. Overall Cost Leadership Requires: Efficient-scale facilities - Overhead control Tight cost - Innovative solution (generally) Seeking out low-cost customers USAA insures only military customers (cost less since they travel a lot hence do business on phone, no need for sales force.) BIM market chain Standardizing a custom service Income tax preparation is a custom service, but H&R Block serves customers requiring routine tax preparation Family health care centers 3-35 Competitive Service Strategies 36 Competitive Service Strategies 1. Overall Cost Leadership Reducing the personnel in service delivery E.g. ATMs Reducing network costs Some service industries require high fixed cost (for network construction) E.g. FedEx use hub-and-spoke for overnight air-package delivery Taking service operations offline Many services must be online, e.g. haircutting If customer need not be present physically, service transaction can be decoupled E.g. American Airlines has one of its 800-number reservations center located in India 3-36 Competitive Service Strategies 37 Competitive Service Strategies 2. Differentiation being unique in brand image, technology use, features, or reputation for customer service. Making the Intangible Tangible (memorable) Hotels providing toiletry items with hotel name affixed Customizing the Standard Product Burger king’s make-to-order effort to differentiate itself from McDonald’s make-to-stock approach hair saloon with personal stylist Reducing Perceived Risk Take extra time to explain the work to be done Village Volvo explain the work and provides service guarantee 3-37 Competitive Service Strategies 38 Competitive Service Strategies 2. Differentiation Giving Attention to Personnel Training results in enhanced service that is difficult to replicate McDonald`s Hamburger University Controlling Quality delivering a consistent level of service Firms have approached this problem in a variety of ways: Personal training, direct supervision, explicit procedures, peer pressure 3-38 Competitive Service Strategies 39 Competitive Service Strategies 3. Focus The idea of servicing a particular target market very well Selected target market: Buyer Group: (e.g. USAA insurance and military officers) Service Offered: (e.g. Motel 6 and budget travelers, FedEx and customers who need overnight package delivery) Geographic Region: (e.g. community college, neighborhood restaurant) 3-39  40 Forecasting Demand For Services  41 Forecasting Models Subjective Models – (initial planning stage) Delphi Methods Cross-Impact Analysis Historical Anology Forecast time horizon Causal Models becomes shorter! Regression Models Time Series Models Moving Averages Exponential Smoothing – (trend and seasonality factors) 14-41  42 Subjective Models initial planning stage few or no data, or insufficient data for long-range forecasts Delphi Methods based on expert opinion. Persons with expertise are asked to make numerical estimates Study is completed in rounds. In each round the results of the previous round are summarized and presented to the experts. The participants are asked to «defend» their answers if they fall outside the interquartile range. A very expensive and time consuming method. Practical only for long term forecasting. 14-42  43 Subjective Models Cross Impact Analysis Assumes that some future event is related to occurence of an earlier event Uses Conditional probability matrix Expert opinion needed Historical Anology Assumes the introduction and growth pattern of a new service will mimic the pattern of a similar concept for which data is available Frequently used to forecast the market penetration or life cycle of a new service. Product life cycle: introduction, growth, maturity, decline e.g. Prediction of color tv market penetration based on experience with black&white tv Anology is questionable 14-43  44 Causal Models Data follow an identifiable pattern over time and an identifiable relationship exists between the information to be forecasted and other factors. Use regression analysis to form a linear relation between independent and dependent variables. Regression Models Regression model is a relationship between the factor being forecasted (dependent variable - 𝑌) and the factors that determine the value of 𝑌 (independent variables - 𝑋𝑖) 𝑌 = 𝑎0 + 𝑎1 𝑋1 + 𝑎2 𝑋2 + ….. + 𝑎! 𝑋𝑛 Econometric Models Econometric models are versions of regression models. consists of a set of simultaneous equations expressing a dependent variable in terms of several independent variables. Require extensive data collection and sophisticated analysis Used for long-range forecasts 14-44  14-45 Causal Models Regression Models – Simple Linear Regression Example: You observed for ten weeks the number of trains produced each week and the total cost of producing those trains. Write a simple linear regression model to estimate the cost of 𝑥 trains. week trains cost 1 10 257.4 2 20 601.6 3 30 782 4 40 765.4 5 45 895.5 6 50 1133 7 60 1152.8 8 55 1132.7 9 70 1459.2 10 40 970  46 Causal Models Regression Models Simple Linear Regression Model Given 𝑛 data points: 𝑥𝑖, 𝑦" , 𝑖 = 1, … , 𝑛 The function that describes 𝑥 and 𝑦 is: 𝑦" = 𝑚𝑥" + 𝑏 The goal is to find the equation of the straight line: 𝑦 = 𝑚𝑥 + 𝑏 ∑$ ∑% ∑ $% Solution: Using 𝑥̅ = , 𝑦1 = , 𝑥𝑦 = ! ! ! 𝑥𝑦 − 𝑥̅ 𝑦1 𝑚= 𝑥 & − 𝑥̅ & 𝑏 = 𝑦1 − 𝑚𝑥̅ 𝑥𝑦 − 𝑥̅ 𝑦1 𝑟= (𝑥 & − 𝑥̅ &)(𝑦 & − 𝑦1 &) 𝑟 is the correlation coefficient a measure of the reliability of the linear relationship between the 𝑥 and 𝑦 values. values close to 1 indicate excellent linear relationship 𝑟 & is the coefficient of determination a measure of the percentage of variation in 𝑦 explained by 𝑥 14-46  14-47 Causal Models Regression Models – Simple Linear Regression Example: Train Cost Model (cont) 𝑥𝑦 − 𝑥̅ 𝑦? 𝑚= 𝑥 ! − 𝑥̅ ! 𝑏 = 𝑦? − 𝑚𝑥̅ 𝑐𝑜𝑠𝑡 = 17,8594 ∗ (𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑡𝑟𝑎𝑖𝑛𝑠) + 164,8651  14-48 Causal Models Regression Models – Simple Linear Regression Example: Train Cost Model 𝑐𝑜𝑠𝑡 = 17,8594 ∗ 𝑡𝑟𝑎𝑖𝑛𝑠 + 164,8651 – How good is that model? !"#!̅ "% Compute 𝑟2 value, recall that: 𝑟 = (!' #!̅ ' )("' #"% ' ) An 𝑟2 value of 0,94 means that the number of trains produced during a week explains 94% of the variation in weekly cost.  49 Time Series Models For making short-term forecasts The values of observations ocur in an identifiable pattern over time 𝑁-Period Moving Average Observations over a period of time appear to have a random pattern To smooth out random variations and produce a reliable estimate of the underlying average Exponential Smoothing – (trend and seasonality factors) 14-49  50 Time Series Models 𝑁-Period Moving Average Let : 𝐴𝑡 = Actual observation for period 𝑡 𝑀𝐴𝑡 = The 𝑁 period moving average at the end of period 𝑡 𝐹()* = Forecast for period 𝑡 + 1 Then: 𝑀𝐴𝑡 = (𝐴( + 𝐴(#* + 𝐴(#+ + ….. +𝐴(#,)* )/𝑁 𝐹()* = 𝑀𝐴( Characteristics: Needs 𝑁 observations to make a forecast Very inexpensive and easy to understand Gives equal weight to all observations Does not consider observations older than 𝑁 periods Smooths out «noise» of occasional blips in the pattern so that we do not overreact to random changes. 14-50  51 Time Series Models 𝑁-Period Moving Average - Example Saturday Occupancy at a 100-room Hotel 3-period Saturday Period Occupancy Moving Avg. Forecast 𝑡 𝐴𝑡 𝑀𝐴𝑡 𝐹( Aug. 1 1 79 Aug. 8 2 84 Aug. 15 3 83 82 Aug. 22 4 81 83 82 Aug. 29 5 98 87 83 Sept. 5 6 100 93 87 Sept. 12 7 93 𝑁=3 Use data of Aug. 1,8,15 to forecast Aug. 22 (𝐹$ ): 𝐹$ = 𝑀𝐴% = (83 + 84 + 79)/3 = 82 14-51  52 Time Series Models Exponential Smoothing Let : 𝑆𝑡 = Smoothed value at end of period 𝑡 𝐴𝑡 = Actual observation for period 𝑡 𝐹()* = Forecast for period 𝑡 + 1 Feedback control nature of exponential smoothing New value (𝑆𝑡 ) = Old value (𝑆(#* ) + 𝛼 [ observed error ] 𝑆( = 𝑆(#* + 𝛼 [𝐴( − 𝑆(#* ] OR: 𝑆( = 𝛼𝐴( + 1 − 𝛼 𝑆(#* 𝐹()* = 𝑆( Exponential smoothing also «smooths out» blips in data but: 1. Old data never dropped out or lost 2. Older data are given progressively less weight 3. Calculation is simple and requires only the most recent data 14-52  53 Time Series Models Exponential Smoothing Weights of Past Demand: Substitute for St = aAt + (1 - a ) St -1 St -1 = aAt + (1 - a )[aAt -1 + (1 - a ) St - 2 ] St = aAt + (1 - a )[aAt -1 + (1 - a ) St - 2 ] St = aAt + a (1 - a ) At -1 + (1 - a ) 2 St - 2 If continued: St = aAt + a (1 - a ) At -1 + a (1 - a ) 2 At - 2 +..... + a (1 - a )t -1 A1 + (1 - a )t S 0 14-53  54 Forecast Error (𝐴! − 𝐹! ) How do we measure the accuracy of these forecasts? Cumulative Forecast Error (CFE) = ∑)&'((𝐴& − 𝐹& ) The sum of forecast errors should tend to zero (summing positive and negative differences). CFE calculates the sum of the forecast errors. ( Mean Absolute Deviation (MAD) = ∑)&'( 𝐴𝑡 − 𝐹𝑡 ) MAD is the most commonly used forecast error. Gives all errors the same weight. ( Mean Squared Error (MSE) = ∑)&'((𝐴& − 𝐹& )! ) MSE squares the errors so large errors have more weight. ( *& +,& Mean Absolute Percentage Error (MAPE) = ∑)&'( (100) ) *! MAPE puts the errors into perspective. An error of 2 is large for a forecast of 10, but insignificant for a forecast of 1000. MAPE enables us see that.  55 Simple Exponential Smoothing Example: Saturday Hotel Occupancy For the first period, smoothed value 𝑆( = 𝐴( (actual value) For 𝑡 = 2: 𝑆! = 𝛼𝐴! + 1 − 𝛼 𝑆( = 81.5 𝐹! = 𝑆( = 79 alfa 0,5 Actual Absolute Squared Occupancy Smoothed Forecast Forecast Error (At- Error Error Percent t (At) Value (St) (Ft) Rounded Ft) (|At-Ft|) (At-Ft)^2 Error 1 79 79 2 84 81,5 79 79 5 5 25 5,95 3 83 82,25 81,5 82 1 1 1 1,20 4 81 81,625 82,25 82 -1 1 1 1,23 5 98 89,8125 81,625 82 16 16 256 16,33 6 100 89,8125 90 10 10 100 10,00 Total 31 33 383 34,72 Mean Mean Mean Absolute Cumulative Absolute Squared Percentage Forecast Forecast Deviation Error Error Error Error (CFE) (MAD) (MSE) (MAPE) 31 6,6 76,6 6,94 14-55  56 Simple Exponential Smoothing Example: Saturday Hotel Occupancy Effect of Alpha (𝛼 = 0.1 vs. 𝛼 = 0.5) 105 100 Actual Occupancy 95 90 Forecast (a = 05.) 85 Forecast 80 (a = 01.) 75 0 1 2 3 4 5 6 Period 14-56  57 Simple Exponential Smoothing Example: Saturday Hotel Occupancy - Compare Errors SimpleExpo SimpleExpo MovingAve alfa=0.5 alfa=0.1 N=3 Cumulative Forecast Error (CFE) 21 45 27 Mean Absolute Deviation (MAD) 5,4 9 9,67 Mean Squared Error (MSE) 64,6 136,6 131,67 Mean Absolute Percentage Error (MAPE) 5,79 9,43 9,85 Notice the error difference with α = 0.5 and α = 0.1. Use Excel solver to select an 𝛼 which minimizes MAD. The value assigned to α is a trade off between overreacting to random fluctuations about a constant mean and detecting a change in mean. Higher α : more responsive to change because of higher weight of recent data. In practice: select an 𝛼 which minimizes MAD  58 Time Series Models Exponential Smoothing With Trend Adjustment What if there is a trend? Trend in a set of data is the average rate at which the observed values change from one period to next over time St = a ( At ) + (1 - a )( St -1 + Tt -1 ) Tt = b ( St - St -1 ) + (1 - b )Tt -1 Ft +1 = St + Tt Trend value 𝑇 is added to the smoothed value 𝑆. Trend is adjusted by a smoothing factor 𝛽 (between 0.1 and 0.5).  59 Time Series Models Exponential Smoothing with Seasonal Adjustment What if there are seasonal effects? (!) Must have actual data for at least one full season. Cycle 𝐿 is the length of one season. 𝐿 : any length of time e.g. 24 hours of a day, 12 months of a year. Remove the seasonality, smooth those data, put the seasonality back and forecast. Seasonality index 𝐼𝑡 – deseasonalize data in a given cycle 𝐿 S t = a ( At / I t - L ) + (1 - a ) S t -1 Ft +1 = ( S t )( I t - L +1 ) At It = g + (1 - g ) I t - L St  60 Time Series Models Exponential Smoothing with Trend and Seasonal Adjustment (Winter’s Model) Is it possible to make more accurate forecasts? Sometimes YES… Include BOTH trend AND Seasonal adjustments by weighting a base smoothed value with trend and seasonal indices. 𝑆& = 𝛼(𝐴& /𝐼&+- ) + (1 − 𝛼) (𝑆&+( + 𝑇&+( ) 𝑇& = 𝛽(𝑆& − 𝑆&+( ) + (1 − 𝛽) 𝑇&+( 𝐹&.( = (𝑆& + 𝑇& ) 𝐼&+-.( For future use update the new season’s 𝐼 as: 𝐼& = 𝛾(𝐴& /𝑆& ) + (1 − 𝛾) 𝐼&+- 14-60  14-61 Other Forecasting Methods Another Time Series Model: ARIMA models – AutoRegressive Integrated Moving Average models An nonseasonal ARIMA model is classified as an ARIMA(𝒑, 𝒅, 𝒒) model. 𝑑 is the number of nonseasonal differences 𝑝 is the number of autoregressive terms (lags of the differenced series) 𝑞 is the number of moving-average terms (lags of the forecast errors) in the prediction equation. A seasonal ARIMA model is classified as an ARIMA(𝒑, 𝒅, 𝒒)𝒙(𝑷, 𝑫, 𝑸) 𝐷 is the number of seasonal differences 𝑃 is the number of seasonal autoregressive terms (lags of the differenced series at multiples of the seasonal period) 𝑄 is the number of seasonal moving average terms (lags of the forecast errors at multiples of the seasonal period).  14-62 Other Forecasting Methods Artificial Intelligence Methods: Artificial neural networks Data mining Machine Learning Pattern Recognition Other methods: Simulation Probabilistic forecasting