Skip to main content
SpringerLink
Log in
Book cover

Essentials of Business Analytics pp 381–420Cite as

Forecasting Analytics

  • Chapter
  • First Online:

  • 194k Accesses

Part of the book series: International Series in Operations Research & Management Science ((ISOR,volume 264))

Abstract

Of course, there is no accurate forecast, but at times this shifts the focus for ... If there is no perfect plan, is there such thing as a good enough plan? …

This is a preview of subscription content, log in via an institution.

Chapter
USD   29.95
Price excludes VAT (USA)
  • Available as PDF
  • Read on any device
  • Instant download
  • Own it forever
eBook
USD   99.00
Price excludes VAT (USA)
  • Available as EPUB and PDF
  • Read on any device
  • Instant download
  • Own it forever
Hardcover Book
USD   129.99
Price excludes VAT (USA)
  • Durable hardcover edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Learn about institutional subscriptions

Notes

  1. 1.

    By Kirk D. Zylstra, 2005, Business & Economics, John Wiley & Sons.

  2. 2.

    Fisher, Anne (May 21, 2013), Big Data could generate millions of new jobs, http://fortune.com/2013/05/21/big-data-could-generate-millions-of-new-jobs [Accessed on Oct 1, 2017].

  3. 3.

    To take into account all available Information that is relevant to the specific forecasting task—usually referred to as Marketing Intelligence.

  4. 4.

    This line is taken from Makridakis et al. (2009).

  5. 5.

    Armstrong, S. (2001), Principles of Forecasting, Kluwer Academic Publishers.

  6. 6.

    From Wikipedia, https://en.wikipedia.org/wiki/Forecasting (accessed on Feb 22, 2018).

  7. 7.

    Among other similar terms like: projecting, extrapolating, foreseeing, etc. In a business context all these terms could be used.

  8. 8.

    Due to limited space, other forecasting methods such as Bayesian forecasting technique, Artificial Neural Network, and special event forecasting are not included.

  9. 9.

    Marketing intelligence or market intelligence—http://en.wikipedia.org/wiki/Market_Intelligence (accessed on Feb 22, 2018).

  10. 10.

    Keast, S., & Towler, M. (2009). Rational Decision-making for Managers: An Introduction (Chapter 2). John Wiley & Sons.

  11. 11.

    The true birth of the Forecasting discipline dates back to late 1970s, early 1980s at the hands of Spyros Makridakis (at INSEAD), Robert Fildes (then at Manchester Business School, now in Lancaster University), and Scott Armstrong (Wharton). Benito Carbone also played a key role in the early stages. The result was to create two journals International Journal of Forecasting—IJF (Elsevier) and Journal of Forecasting—JoF (Wiley), a conference ISF (https://isf.forecasters.org/, accessed on Feb 22, 2018), an Institute IIF (www.forecasters.org, accessed on Feb 22, 2018), in a word ... a DISCIPLINE! Many have followed since then and are part of the forecasting community now, including the authors of these texts, but history was written by those 3–4 men and their close associates. More details can be found in the interview of Spyros for IJF: Fildes, R. and Nikolopoulos, K. (2006) “Spyros Makridakis: An Interview with the International Journal of Forecasting”. International Journal of Forecasting, 22(3): 625–636.

  12. 12.

    https://link.springer.com/chapter/10.1007/978-3-642-25646-2_56 accessed on Sep 11, 2017.

  13. 13.

    For this and other types of forecast classifications, see Hibon and Makridakis (2000).

  14. 14.

    A standard engineering expression, for a situation or a solution where something seems to work fine, but we are not sure why and definitely do not know for how long it will keep on working!!

  15. 15.

    For more information please visit http://www.forecastingprinciples.com/, or read “Principles of Forecasting: A Handbook for Researchers and Practitioners, J. Scott Armstrong (ed.): Norwell, MA: Kluwer Academic Publishers, 2001”.

  16. 16.

    A term often used in engineering applications.

  17. 17.

    Of course there would still be some noise over this line.

  18. 18.

    Usually increasing the forecasts, due to an optimism bias (more on this and other types of bias in Chap. 13).

  19. 19.

    Noise is a term met in many sciences. I prefer the electrical engineering definition of it where Noise can block, distort, or change/interfere with the meaning of a message in both human and electronic communication.

  20. 20.

    Armstrong 2001.

  21. 21.

    Timmerman and Granger 2004.

  22. 22.

    Syntetos et al. 2010.

  23. 23.

    Maris et al. 2007; Bozos et al. 2008.

  24. 24.

    www.forecastingprinciples.com [Accessed on Oct 1, 2017].

  25. 25.

    Professor J. Scott Armstrong, http://www.jscottarmstrong.com/ [Accessed on Oct 1, 2017].

  26. 26.

    International Institute of Forecasters, https://forecasters.org/ [Accessed on Oct 1, 2017].

  27. 27.

    A famous quote attributed to Albert Einstein.

References

  • Andrawis, R. R., Atiya, A. F., & El-Shishiny, H. (2011). Forecast combinations of computational intelligence and linear models for the NN5 time series forecasting competition. International Journal of Forecasting, 27, 672–688.

    Article  Google Scholar 

  • Armstrong, J. S. (2001). Principles of forecasting: A handbook for researchers and practitioners. Dordrecht: Kluwer Academic Publishers.

    Book  Google Scholar 

  • Assimakopoulos, V., & Nikolopoulos, K. (2000). The theta model: A decomposition approach to forecasting. International Journal of Forecasting, 16, 521–530.

    Article  Google Scholar 

  • Babai, M. Z., Ali, M., & Nikolopoulos, K. (2012). Impact of temporal aggregation on stock control performance of intermittent demand estimators: Empirical analysis. OMEGA: The International Journal of Management Science, 40, 713–721.

    Article  Google Scholar 

  • Bollerslev, T., Engle, R. F., & Nelson, D. B. (1994). ARCH models. In R. F. Engle & D. L. McFadden (Eds.), Handbook of econometrics (Vol. 4, pp. 2959–3038). Amsterdam: North-Holland.

    Google Scholar 

  • Box, G. E. P., & Jenkins, G. M. (1970). Time series analysis: Forecasting and control. San Francisco, Holden Day (revised ed. 1976).

    Google Scholar 

  • Bozos, K., Nikolopoulos, K., & Bougioukos, N. (2008). Forecasting the value effect of seasoned equity offering announcements. In 28th international symposium on forecasting ISF 2008, June 22–25 2008. France: Nice.

    Google Scholar 

  • Brown, R. G. (1956). Exponential smoothing for predicting demand. Cambridge, MA: Arthur D. Little Inc.

    Google Scholar 

  • Chatfield, C. (2005). Time-series forecasting. Significance, 2(3), 131–133.

    Article  Google Scholar 

  • Croston, J. D. (1972). Forecasting and stock control for intermittent demands. Operational Research Quarterly, 23, 289–303.

    Article  Google Scholar 

  • Efron, B. (1979). Bootstrap methods: Another look at the jackknife. The Annals of Statistics, 7, 126.

    Article  Google Scholar 

  • Engle, R. F. (1982). Autoregressive conditional heteroscedasticity with estimates of the variance of the United Kingdom inflation. Econometrica, 50, 987–1008.

    Article  Google Scholar 

  • Goodwin, P. (2006). Decision Analysis for Management Judgement, 3rd Edition Chichester: Wiley.

    Google Scholar 

  • Hanke, J. E., & Wichern, D. W. (2005). Business forecasting (8th ed.). Upper Saddle River: Pearson.

    Google Scholar 

  • Harrison, P. J., & Stevens, C. F. (1976). Bayesian forecasting. Journal of the Royal Statistical Society (B), 38, 205–247.

    Google Scholar 

  • Hibon, M., & Makridakis, S. (2000). The M3 competition: Results, conclusions and implications. International Journal of Forecasting, 16, 451–476.

    Article  Google Scholar 

  • Holt, C. C. (1957). Forecasting seasonals and trends by exponentially weighted averages. O. N. R. Memorandum 52/1957. Pittsburgh: Carnegie Institute of Technology. Reprinted with discussion in 2004. International Journal of Forecasting, 20, 5–13.

    Article  Google Scholar 

  • Hyndman, R. J. (2014). Forecasting – Principle and practices. University of Western Australia. Retrieved July 24, 2017, from robjhyndman.com/uwa.

  • Johnston, F. R., Boylan, J. E., & Shale, E. A. (2003). An examination of the size of orders from customers, their characterization and the implications for inventory control of slow moving items. Journal of the Operational Research Society, 54(8), 833–837.

    Article  Google Scholar 

  • Jose, V. R. R., & Winkler, R. L. (2008). Simple robust averages of forecasts: Some empirical results. International Journal of Forecasting, 24(1), 163–169.

    Google Scholar 

  • Keast, S., & Towler, M. (2009). Rational decision-making for managers: An introduction. Hoboken, NJ: John Wiley & Sons.

    Google Scholar 

  • Kourentzes, N. (2014). Improving your forecast using multiple temporal aggregation. Retrieved August 7, 2017, from http://kourentzes.com/forecasting/2014/05/26/improving-forecasting-via-multiple-temporal-aggregation.

  • Kourentzes, N., Petropoulos, F., & Trapero, J. R. (2014). Improving forecasting by estimating time series structural components across multiple frequencies. International Journal of Forecasting, 30, 291–302.

    Article  Google Scholar 

  • Leven and Segerstedt. (2004). Referred to in Syntetos and Boylan approximation section.

    Google Scholar 

  • Lindsey, M., & Pavur, R. (2008). A comparison of methods for forecasting intermittent demand with increasing or decreasing probability of demand occurrences. In K. D. Lawrence & M. D. Geurts (Eds.), Advances in business and management forecasting (advances in business and management forecasting) (Vol. 5, pp. 115–132). Bingley, UK: Emerald Group Publishing Limited.

    Chapter  Google Scholar 

  • Makridakis, S., Hogarth, R., & Gaba, A. (2009). Dance with chance: Making luck work for you. London, UK: Oneworld Publications.

    Google Scholar 

  • Makridakis, S., Wheelwright, S. C., & Hyndman, R. J. (1998). Forecasting: Methods and applications (3rd ed.). New York: John Wiley and Sons.

    Google Scholar 

  • Maris, K., Nikolopoulos, K., Giannelos, K., & Assimakopoulos, V. (2007). Options trading driven by volatility directional accuracy. Applied Economics, 39(2), 253–260.

    Article  Google Scholar 

  • Nikolopoulos, K., Assimakopoulos, V., Bougioukos, N., Litsa, A., & Petropoulos, F. (2011a). The theta model: An essential forecasting tool for supply chain planning. Advances in Automation and Robotics, 2, 431–437.

    Article  Google Scholar 

  • Nikolopoulos, K., Syntetos, A., Boylan, J., Petropoulos, F., & Assimakopoulos, V. (2011b). ADIDA: An aggregate/disaggregate approach for intermittent demand forecasting. Journal of the Operational Research Society, 62, 544–554.

    Article  Google Scholar 

  • Petropoulos, F., Makridakis, S., Assimakopoulos, V., & Nikolopoulos, K. (2014). ‘Horses for Courses’ in demand forecasting. European Journal of Operational Research, 237, 152–163.

    Article  Google Scholar 

  • Prestwich, S. D., Tarim, S. A., Rossi, R., & Hnich, B. (2014). Forecasting intermittent demand by hyperbolic-exponential smoothing. International Journal of Forecasting, 30(4), 928–933.

    Article  Google Scholar 

  • Reeb, D. M., Sakakibara, M., & Mahmood, I. P. (2012). From the editors: Endogeneity in international business research. Journal of International Business Studies, 43, 211–218.

    Article  Google Scholar 

  • Rostami-Tabar, B., Babai, M. Z., Syntetos, A. A., & Ducq, Y. (2013). Demand forecasting by temporal aggregation. Naval Research Logistics, 60, 479–498.

    Article  Google Scholar 

  • Spithourakis, G. P., Petropoulos, F., Babai, M. Z., Nikolopoulos, K., & Assimakopoulos, V. (2011). Improving the performance of popular supply chain forecasting techniques: An empirical investigation. Supply Chain Forum: An International Journal, 12, 16–25.

    Article  Google Scholar 

  • Syntetos, A. A., & Boylan, J. E. (2001). On the bias of intermittent demand estimates. International Journal of Production Economics, 71, 457–466.

    Article  Google Scholar 

  • Syntetos, A. A., & Boylan, J. E. (2005). The accuracy of intermittent demand estimates. International Journal of Forecasting, 21, 303–314.

    Article  Google Scholar 

  • Syntetos, A. A., Nikolopoulos, K., & Boylan, J. E. (2010). Judging the judges through accuracy-implication metrics: The case of inventory forecasting. International Journal of Forecasting, 26, 134–143.

    Article  Google Scholar 

  • Taylor, A. R. (1997). On the practical problems of computing seasonal unit root tests. International Journal of Forecasting, 13(3), 307–318.

    Article  Google Scholar 

  • Teunter, R. H., Syntetos, A., & Babai, Z. (2011). Intermittent demand: Linking forecasting to inventory obsolescence. European Journal of Operational Research, 214, 606–615.

    Article  Google Scholar 

  • Thomakos, D. D., & Nikolopoulos, K. (2014). Fathoming the theta method for a unit root process. IMA Journal of Management Mathematics, 25, 105–124.

    Article  Google Scholar 

  • Timmerman, A., & Granger, C. W. J. (2004). Efficient market hypothesis and forecasting. International Journal of Forecasting, 20, 15–27.

    Article  Google Scholar 

  • Tseng, F., Yu, H., & Tzeng, G. (2002). Combining neural network model with seasonal time series ARIMA model. Technological Forecasting and Social Change, 69, 71–87.

    Article  Google Scholar 

  • Willemain, T. R., Smart, C. N., & Schwarz, H. F. (2004). A new approach to forecasting intermittent demand for service parts inventories. International Journal of Forecasting, 20, 375–387.

    Article  Google Scholar 

  • Wright, G., & Goodwin, P. (1998). Forecasting with judgement. Chichester and New York: John Wiley and Sons.

    Google Scholar 

Download references

Author information

Authors and Affiliations

  1. Bangor Business School, Bangor, Gwynedd, UK

    Konstantinos I. Nikolopoulos

  2. University of Peloponnese, Tripoli, Greece

    Dimitrios D. Thomakos

Authors
  1. Konstantinos I. Nikolopoulos
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Dimitrios D. Thomakos
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Konstantinos I. Nikolopoulos .

Editor information

Editors and Affiliations

  1. Applied Statistics and Computing Lab, Indian School of Business, Hyderabad, Telangana, India

    Bhimasankaram Pochiraju

  2. Gies College of Business, University of Illinois at Urbana Champaign, Champaign, IL, USA

    Sridhar Seshadri

1 Electronic Supplementary Material

Supplementary Data 12.1

Data—ARIMA (CSV 382 bytes)

Supplementary Data 12.2

Data—Croston and SBA (CSV 92 bytes)

Supplementary Data 12.3

Data—Damped Holt (CSV 121 bytes)

Supplementary Data 12.4

Data—SES, ARRSES, Holt, HoltWinter (CSV 440 bytes)

Supplementary Data 12.5

Data—Theta (CSV 241 bytes)

Supplementary Data 12.6

FA—Excel Template (XLSX 64 kb)

Supplementary Data 12.7

Forecasting Analytics (R 2 kb)

Supplementary Data 12.8

Forecasting chapter—Consolidated Output v1.0 2017-11-21 (XLSX 27 kb)

Appendix 1

Appendix 1

Example: The monthly sales (in million USD) of Vodka is given for the period 1968–1970. We want to forecast the sales for the year 2016 using various forecasting methods—SES, ARRSES, Holt, Holt–Winters (Additive/Multiplicative).

Data: The data can be downloaded from the book’s website and the dataset name is “Data - SES, ARRSES, Holt, HoltWinter.csv”. You can also refer to Table 12.6 for data.

Table 12.6 Data for SES, Holt, ARRSES, and Holt–Winters method
Full size table

R Code (to read data)

read.csv (“filename.ext”, header = TRUE)

SES Method

Install forecast package

install.packages(“forecast”)

R function

ses (<Univariate vector of observations>, h = <number of periods to forecast)

Note: The ses function in R by default optimizes both the value of alpha and the initial value.

In case you prefer the output for a specified alpha value then use parameter

<initial = ”simple”>

and set the alpha value in the parameters.

ses(<univariate vector of observations>, h = <number of periods to forecast>, alpha = < >, initial = “simple”)

The above code will set the first forecast value equal to first observation. If alpha is omitted it will optimize for alpha.

Holt Method

Install forecast package

install.packages(“forecast”)

R function.

holt (<univariate vector of observations>, h = <number of periods to forecast>)

Note: The holt function by default optimizes both the value of alpha and the initial value.

In case you prefer the output for a specified alpha and beta value then use

<initial = ”simple”>

parameter and set the alpha and beta values in the parameters.

holt (<univariate vector of observations>, h = <number of periods to forecast>, alpha = < >, beta = < >, initial = “simple”)

The above code sets first level equal to first value and trend as difference of first two values. If alpha is omitted it will optimize for alpha.

Holt–Winters Method

Install stats package

install.packages(“stats”)

R function

HoltWinters (<name of dataset>, alpha = <>, beta = <>, gamma = <>, seasonal = c("additive”, "multiplicative"), start.periods = 2, l.start = NULL, b.start = NULL, s.start = NULL, optim.start = c(alpha = 0.3, beta = 0.1, gamma = 0.1), optim.control = list())

The value of alpha, beta, and gamma can be either initialized by specifying <alpha>, <beta>, <gamma> and if they are NULL it will optimize the values as specified in optim.start. You can also specify starting values of alpha, beta, and gamma to optimize using <optim.start> parameter. Seasonality can be considered additive or multiplicative. The <start.periods> is the initial data used to start the forecast (minimum 2 seasons of data). Starting values of level <l.start>, trend <b.start>, and seasonality <s.start> can be either be initialized or optimized by setting equal to NULL.

For the HoltWinters function, the dataset must be defined as a time-series (ts) type. A dataset can be converted to time-series type, using the below code:

ts (<name of dataset>, frequency = number of periods in a season)

Damped Holt Method

Data: The data can be downloaded from the book’s website and the dataset name is “Data - Damped Holt.csv”. You can also refer to Table 12.7 for data.

Table 12.7 Data for damped exponential smoothing using Holt’s method
Full size table

Install forecast package

install.packages(“forecast”)

R function

holt (<univariate vector of observations>, h = <number of periods to forecast>, damped = TRUE)

Note: The holt function by default optimizes both the value of alpha and the initial value.

In case you prefer the output for a specified alpha, beta, and phi values then use <initial = ”simple”> parameter and set the alpha, beta, and phi values in the parameters.

holt (<univariate vector of observations>, h = <number of periods to forecast>, damped = TRUE, alpha = < >, beta = < >, phi = < >)

If alpha, beta, and phi are omitted, it will optimize for these values.

Theta method

Data: The data can be downloaded from the book’s website and the dataset name is “Data - Theta.csv”. You can also refer to Table 12.8 for data.

Table 12.8 Data for Theta model
Full size table

Install forecTheta package

Install.packages(“forectheta”)

R function

stm (ts (<univariate vector of observations>), h = <number of periods to forecast>, par_ini = c (y[1]/2, 0.5,2))

Refer https://cran.r-project.org/web/packages/forecTheta/forecTheta.pdf for more details.

Note: You may try either “stm” or “stheta.” There is a slight difference in the implementation of the original method.

ARIMA method

Data: The data can be downloaded from the book’s website and the dataset name is “Data - ARIMA.csv”.

Install forecast package

install.packages(“forecast”)

R function

arima (ts (<univariate vector of observations>, freq = <period of data>), order = c(<p>,<d>,<q>))

To view the fitted coefficients, store the output and call that array.

To forecast, use the command:

forecast (<name of output>, h = <number of periods to forecast>)

Croston and SBA method

Data: The data can be downloaded from the book’s website and the dataset name is “Data - Croston and SBA.csv”. You can also refer to Table 12.9 for data.

Table 12.9 Data for Croston and SBA model
Full size table

Install tsintermittent package

install.packages(“tsintermittent”)

R function

crost (ts(<univariate vector of observations>), h = <number of periods to forecast>, w = c(<>,<>), init = c(<>,<>), type = “croston”, init.opt = FALSE)

Refer https://cran.r-project.org/web/packages/tsintermittent/tsintermittent.pdf for more details.

<crost> function operates on the time-series vector. Initial values can be either differently chosen or provided as a vector of demand and interval value. <type> refers to the model used. Cost to the optimization criterion. If <init.opt> is TRUE, it will optimize the initial values. If <w> is NULL, it will optimize the smoothing parameters.

Consolidated Forecast Output for Vodka Example

See Tables 12.10, 12.11, 12.12, 12.13, and 12.14.

Table 12.10 Consolidated output of SES, ARRSES, Holt, Holt–Winters methods (*R Output, ^Excel Output)
Full size table
Table 12.11 Forecast using Damped Holt method
Full size table
Table 12.12 Forecast using Theta method
Full size table
Table 12.13 Forecast using ARIMA method
Full size table
Table 12.14 Forecast using Croston and SBA methods
Full size table

Rights and permissions

Reprints and permissions

Copyright information

© 2019 Springer Nature Switzerland AG

About this chapter

Check for updates. Verify currency and authenticity via CrossMark

Cite this chapter

Nikolopoulos, K.I., Thomakos, D.D. (2019). Forecasting Analytics. In: Pochiraju, B., Seshadri, S. (eds) Essentials of Business Analytics. International Series in Operations Research & Management Science, vol 264. Springer, Cham. https://doi.org/10.1007/978-3-319-68837-4_12

Download citation

Publish with us

Policies and ethics

Search

Navigation