8.2 A Forecaster Toolkit - A Tool for Every Forecasting Setting


  1. ^ Statistical Methods
  2. ^ Multiple Regression
  3. ^ Michelin model of forecasting demand for the replacement automobile tire market
  4. ^ calibrating of the statistical model
  5. ^ limitations of the statistical methods
  6. ^ the future will look very much like the past
  7. ^ US West - the regional telephone company case
  8. ^ statistical models used to predict needs for telephone capacity failed to allow for rapidly increasing use of computer modems, faxes
  9. ^ Type your reference here.
  10. ^ Tyre Makers
  11. ^ the difussion of innovation process
  12. ^ the impact on consumer demand of different combinations of attributes

[1] Statistical methods use past history and various statistical techniques, such as[2] multiple regression or time series analysis, to forecast the future based on anextrapolation of the past.

This method is typically not useful for ACG or other entrepreneurs or new product managers charged with forecasting sales for a new product or new business. There is no history in their venture on which to base a statistical forecast. In established firms, for established products, statistical methods are extremely useful.

[3] When Michelin, the tire maker, wants to forecast demand for the replacement automobile tire market in Asia for the next year, it can build a statistical model using such factors as the number and age of vehicles currently on the road in Asia, predictions of GDP for the region, the last few years’ demand, and other relevant factors to forecast market potential as well as Michelin’s own replacement tire sales for the coming year.

Such a procedure is likely to result in a more accurate forecast than other methods, especially if Michelin has years of experience with which to [4] calibrate its statistical model.

As with all forecasting methods, statistical methods [5] have important limitations.
1 . Most important of these is that statistical methods generally assume that [6] .

Sometimes this is not the case.

Example :

[7] US WEST, the regional Bell telephone company serving the Rocky Mountain and Northwest regions of the United States, ran into trouble in the 1990s when its[8] statistical models used to predict needs for telephone capacity failed to allow for rapidly increasing use of computer modems, faxes, and second lines for teenagers in American homes.Suddenly, the average number of lines per home skyrocketed, and there was not enough physical plant – cable in the ground, switches, and so on – to accommodate the growing demand. Consumers had to wait, sometimes for months, to get additional lines, and they were not happy about it!

2. Similarly, if product or market characteristics change,[9] statistical models used without adequate judgement may not keep pace.

Example : [10] Tire Makers

When tire makers produce automobile tires that last 80 000 miles insteadof 30 000 to 50 000 miles, the annual demand for replacement tires is reduced.
If automobile manufacturers were to change the number of wheels on the typical car from four, the old statistical models would also be in trouble.

Other quantitative forecasting methods, especially for new product forecasting, have also been developed.

These include methods to mathematically model [11] the diffusion of innovation process for consumer durables (discussed in Module 5) and conjoint analysis, a method to forecast [12] the impact on consumer demand of different combinations of attributes that might be included in a new product

  1. client contact systems
  2. collector bias
  3. competitive advantage
  4. competitive intelligence
  5. computerised reorder system
  6. consumer behaviour
  7. data sources
  8. evidence based forecast
  9. experienced user
  10. internal records
  11. just in time
  12. logistical alliance
  13. market potential
  14. market segmentation
  15. market segments
  16. marketing program
  17. marketing research
  18. mass market
  19. mass market strategy
  20. michelin; us west;
  21. micro segmentation
  22. middleman
  23. modified rebuy
  24. multi-functional sales teams
  25. multilevel selling
  26. multiple buying
  27. multiple level relationships
  28. mutual trust
  29. narrow market segment
  30. narrow niche
  31. nationalisation of producers
  32. nerve center
  33. new task buy
  34. nine west group
  35. observation;direct observation' tanzania mobile;
  36. on-time delivery
  37. opportunity; research
  38. order handling
  39. organisation market
  40. organization marketing behaviour
  41. organizational behaviour
  42. organizational customers
  43. organizational demand
  44. organizational market
  45. organizational purchasing behaviour
  46. organizational purchasing process
  47. paperless exchange
  48. parity pricing
  49. personal selling
  50. personal use
  51. political risk
  52. potential market; penetrated market
  53. pre-delivery inspection
  54. pre-sale service
  55. prestige buyer
  56. pretender
  57. primary data
  58. procurement costs
  59. purchasing criteria
  60. qualitative data
  61. qualitative research
  62. quality assurance
  63. quality standards
  64. quantitative data
  65. quantitative research
  66. research objectives
  67. retention programme
  68. routine purchase
  69. sales forecast
  70. semantic differentiation scale
  71. sequence of information
  72. shared costs
  73. short term contracts
  74. social construction
  75. status oriented consumers
  76. stock availability
  77. straight rebuy
  78. supplier bargaining power
  79. supplier performance
  80. supplier reputation
  81. survey
  82. tabulation errors
  83. tanzania mobile
  84. target customers
  85. target market
  86. target marketing
  87. technical experts;
  88. test markets
  89. transaction cost
  90. trend forecasting
  91. trusting patron
  92. underlying consumer demand
  93. unethical demands
  94. unstated but implicit assumptions
  95. users
  96. value analysis
  97. value shopper
  98. vertical integration
  99. visceral thing that cannot be trained
  100. wild guess