Module 6 Forecasting - Lecture notes 6 PDF

Preview

Summary

Module on forecasting...

Description

OBJECTIVES

At the completion of this module, you will be able to: ● ● ● ● ● ● ● ●

Explain the nature and importance of forecasting. Explain the types of forecasting and correlation analysis models. Explain the components of time series. Apply moving average forecasting models. Apply simple exponential smoothing forecasting. Apply trend projection forecasting. Forecast using data with seasonality Compute measures of forecast accuracy.

Forecasting ● Process of predicting future events ○ Underlying basis of business decisions (production, inventory, personnel, facilities) ○ Importance ■ Supply chain ● Good supplier relations, innovation, cost/speed to market ■ Quality management ● Forecasting customer demand accurately ■ Human resources ● Hiring, training, laying off workers ■ Capacity ● Shortages cause undependable delivery, loss of customers & market share ■ Strategic planning ● Requires accurate forecasts of future products/markets ○ In service sector: presents unusual challenges ■ Special need for short-term records ■ Holidays & other unusual events ■ Needs differ through industries and products ● Forecasting Time Horizons ○ Short range: up to 1 year ■ purchasing , job scheduling, workforce levels, job assignments, production levels ■ Operational decision-making contexts

○ Medium range: 1-3 years ■ Sales & production planning, budgeting ■ tactical/managerial decision-making context ○ Long range: 3+ years ■ New product planning, facility location, R&D ■ Strategic decision making ○ Influence on product life cycle ■ Intro/growth need longer forecasts than maturity/decline ■ Project staffing, inventory, & factory levels ● Types of Forecasts ○ Economic forecasts ■ address business cycle, inflation, money supply, housing starts, etc. ○ Technological forecasts ■ Predict rate of technological progress, impacts development of new products ○ Demand forecasts ■ Predict sales of new & existing products Forecasting Steps 1. Determine use of forecast 2. Select items to be forecasted 3. Determine time horizon of forecast 4. Select forecasting model(s) 5. Gather data needed 6. Make forecast 7. Validate & implement results Forecasting Approaches ● Qualitative Methods: Used when situation vague and little data exists ■ Involves intuition & experience ■ New products or technology ○ Methods ■ Jury of executive opinion ● Small group of High level experts/managers ● Combines with some statistical methods ● Relatively quick, “group-think” advantage

■ Sales force composite ● Estimates from individual salespersons projected sales ● reviewed for reasonableness, then aggregated at national/district levels ● May be optimistic ■ Market survey ● Ask the customer about purchasing plans ● Use for demand, planning, product design ● May be optimistic and not be what customer actually does ■ Delphi method ● Panel of experts, queried iteratively ● Staff administers survey to respondents, decision makers evaluate to make decisions ● Quantitative methods ○ Stable situations where historical data exists ■ Existing products, current technology ○ Involves mathematical techniques ○ Methods ■ Naive approach (time-series model) ● Issues demand same in next & most recent period ● Sometimes cost effective & efficient ● Can be good starting point, but not good used by itself ■ Moving average(time series model) ● Simple moving average ○ If little or no trend, used often for smoothing

● Weighted Moving Average ○ Trends, prioritizes newer data ○ Weights based on experience/intuition ○ ● Issues with moving averages ○ Increasing n makes less sensitive to changes

○ Doesn’t forecast well ○ Requires extensive historical data ■ Exponential smoothing (time series model) ● Form of weighted moving average ○ Weights decline exponentially, most recent data weighted most ● Requires little record keeping ● Smoothing constant subjectively chosen between 0 & 1 (generally between 0.05-0.5) ○ Closer to 0, extreme stability, chosen when average stable ○ Closer to 1 when average likely to change

○ Smoothing Constant methods ■ Mean Absolute Deviation

■ Mean Squared Error

■ Mean absolute Percent Error

■ Adaptive smoothing ● Computer continually monitors error and adjusts values ■ Trend progression(time series model) ● Historical data to project medium to long term ● Uses least squares

○ Requirements of LSM ■ Always plot data to ensure linear relationship

■ Do not predict time period far beyond database ■ Deviations assumed to be random ■ Linear regression (associative model) ○ Time series model

■ Evenly spaced numerical data from past ■ Components ● Trend ○ Overall upward or downward pattern ○ Changes due to population, age, culture, tech., etc ○ Typically several years duration ● Seasonal ○ Regular pattern of up/down fluctuations ○ Occurs within single year ■ Due to weather, customs, etc. ○ Multiplicative Seasonal Model ■ Compute seasonal indices ■ Deseasonalize data ■ Compute the deseasonalized forecasts using moving averages, exponential smoothing, or trend projection

■ Apply seasonal indices to deseasonalized forecasts ● Cyclical ○ Repeating up & down movements ■ Affected by business, political, economic factors ○ Multiple years duration ○ Not observable for short term ○ Causal or associative relationships ● Random ○ Also called “error” ■ Erratic & unsystematic ○ Short duration, non repeating ○ Due to unforeseen events & random variation ○ Correlation ■ Strength of linear relationship between variables ■ Does not imply causality ■ Coefficient of correlation: -1 to 1

○ Coefficient of Determination: r2 ■ Measures percent change in y predicted by change in x ■ 0-1 ○ Tracking Signal ■ How well forecast products absolute values

● Good tracking signal has low values ● If forecasts continually high/how, has bias error

Video: Hard Rock Cafe ● Forecasting fundamental for all organization ○ Expansion needs, demand, profits, how many to hire ○ Service ■ Forecast capacity for long term at each location (55 cafes, monthly forecast at each) ■ Purchasing forecast blanket orders for medium term

● ex) leather jackets 8 months out ■ Short term scheduling & purchasing products, labor, & supplies ● Food, labor ● Sales forecast ○ Plug in all members of organization ■ Purchasing managers beef, chicken , pork long time contracts based on forecast ■ New locations and customers considered ■ Sales drives profit drives cash flow forecast ● Need for borrowing needs at banks ● Moving averages ● Weighted moving averages ○ 3 year for manager bonus targets ■ 40:40:20 in order of recent years ■ Bonus if exceed target ● Exponential smoothing ● Regression analysis ○ For menu changes ○ Primary issue is number of people walking through door ■ POS equipment uses number of entrees and physical human being count as well ● Total vs. how many ordering entrees ○ Sent to data warehouse and compile statistics on average consumer ○ Hw reflects remodeling and menu changes as well...