Abstract
Market researchers often conduct surveys to measure how much value consumers place on the various features of a product. The resulting data should enable managers to combine these utility values in different ways to predict the market share of a product with a new configuration of features. Researchers assess the accuracy of these choice models by measuring the extent to which the summed utilities can predict actual market shares when respondents choose from sets of complete products. The current paper includes data from 201 consumers who gave ratings to 18 cell phone features and then ranked eight complete cell phones. A simple summing of the utility values predicted the correct product on the ranking task for 22.8 % of respondents. Another accuracy measurement is to compare the market shares for each product using the ranking task and the estimated market shares based on summed utilities. This produced a mean absolute difference between ranked and estimated market shares of 7.8 %. The current paper applied two broad strategies to improve these prediction methods. Various evolutionary search methods were used to classify the data for each respondent to predict one of eight discrete choices. The fitness measure of the classification approach seeks to reduce the Classification Error Percent (CEP) which minimizes the percent of incorrect classifications. This produced a significantly better fit with the hit rate rising from 22.8 to 35.8 %. The mean absolute deviation between actual and estimated market shares declined from 7.8 to 6.1 % (p. <0.01). A simple language specification will be illustrated to define symbolic regression and classification searches.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Breiman L, Friedman J, RAOlshen, Stone C (1984) Classification and regression trees. Wadsworth and Brooks, Pacific Grove
Codd EF (1983) A relational model of data for large shared data banks. Commun ACM 26(1): 64–69
Green PE, Rao V (1971) Conjoint measurement for quantifying judgmental data. J Mark Res 8(3):355–363
Korns MF (2007) Large-scale, time-constrained symbolic regression-classification. In: Riolo RL, Soule T, Worzel B (eds) Genetic programming theory and practice V. Genetic and evolutionary computation, chap 4. Springer, Ann Arbor, pp 53–68. doi:10.1007/978-0-387-76308-8_4
Korns MF (2010) Abstract expression grammar symbolic regression. In: Riolo R, McConaghy T, Vladislavleva E (eds) Genetic programming theory and practice VIII. Genetic and Evolutionary Computation, vol 8, chap 7. Springer, Ann Arbor, pp 109–128. http://www.springer.com/computer/ai/book/978-1-4419-7746-5
Korns MF (2011) Accuracy in symbolic regression. In: Riolo R, Vladislavleva E, Moore JH (eds) Genetic programming theory and practice IX. Genetic and evolutionary computation, chap 8. Springer, Ann Arbor, pp 129–151. doi:10.1007/978-1-4614-1770-5_8
Marder E (1997) The laws of choice: predicting customer behavior. The Free Press, New York
McCulloch W, Pitts W (1943) A logical calculus of the ideas immanent in nervous activity. Bull Math Biophys 5(4):115–133
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendices
Appendix 1: Questionnaire Text
-
1.
Operating system
-
a.
Android
-
b.
Symbian
-
c.
Windows
-
d.
Blackberry
-
e.
iOS (iPhone OS)
-
a.
-
2.
Screen size
-
a.
Less than 3 in.
-
b.
3.0–3.4 in.
-
c.
3.5–3.9 in.
-
d.
4.0–4.4 in.
-
e.
4.5–4.9 in.
-
f.
5 in. and over
-
a.
-
3.
Camera memory
-
a.
Below 2 megapixels
-
b.
2–4.9 megapixels
-
c.
5–7.9 megapixels
-
d.
8 Megapixels and above
-
a.
-
4.
Memory
-
a.
Below 8 GB
-
b.
8–15.9 GB
-
c.
16–31.9 GB
-
d.
32–63.9 GB
-
e.
64 GB or more
-
a.
-
5.
Talk time
-
a.
Less than 6 h
-
b.
6–11 h
-
c.
12–23 h
-
d.
24–35 h
-
e.
36 h or more
-
a.
-
6.
Stand by time
-
a.
Under 50 h
-
b.
50–99 h
-
c.
100–199 h
-
d.
200–299 h
-
e.
300 h or more
-
a.
-
7.
Price
-
a.
5000 Rs or less
-
b.
5001–10,000 Rs
-
c.
10,001–18,000 Rs
-
d.
18,001–35,000 Rs
-
e.
35,001 Rs and above
-
a.
-
8.
Phone thickness
-
a.
Less than 6 mm
-
b.
6–7 mm
-
c.
8–9 mm
-
d.
10–11 mm
-
e.
12 mm or more
-
a.
-
9.
CPU speed
-
a.
1 GHz or less
-
b.
1.0–1.3 GHz
-
c.
1.4–1.5 GHz
-
d.
1.6–1.9 GHz
-
e.
2.0 GHz or more
-
a.
-
10.
Warranty length
-
a.
Free repairs for 6 months
-
b.
Free repairs for 1 year
-
c.
Free repairs for 1.5 years
-
d.
Free repairs for 2 years
-
e.
Free repairs for 2.5 years
-
a.
-
11.
GPS
-
a.
Has GPS
-
b.
No GPS
-
a.
-
12.
Wi-Fi
-
a.
Has Wi-Fi
-
b.
No Wi-Fi
-
a.
-
13.
Touchscreen
-
a.
Has a touchscreen
-
b.
No touchscreen
-
a.
-
14.
SIM format
-
a.
Single SIM
-
b.
Dual SIM
-
a.
-
15.
3G
-
a.
Has 3G connectivity
-
b.
No 3G connectivity
-
a.
-
16.
Qwerty keyboard
-
a.
Has a QWERTY keyboard
-
b.
No QWERTY keyboard
-
a.
-
17.
Brand impression
-
a.
Apple
-
b.
Samsung
-
c.
Blackberry
-
d.
XOLO
-
e.
Spice
-
f.
Micromax
-
g.
Nokia
-
h.
Lava
-
a.
Appendix 2: Sources of Feature Data
All feature data for the eight mobile phones were drawn from www.Flipkart.com on September 26th, 2013 except the following items that were missing from the Flipkart comparison screens.
For the iPhone 5 with 32 GB, data was missing for the CPU speed attribute. This was taken from www.GSMArena.com on September 26th 2013.
For the Samsung Galaxy Note 2, data was missing for the talk-time and standby time attributes. This was taken from www.GSMArena.com on September 26th 2013.
For the Blackberry Curve 9220, data was missing for the GPS attribute. This was taken from www.GSMArena.com on September 26th 2013. The CPU speed attribute was missing from both these sources. It was taken from asia.cnet.com on September 26th 2013.
For the XOLO Q1000, data was missing for the GPS attribute. This was taken from www.GSMArena.com on September 26th 2013.
For the Spice MI-495, data was missing for the USB connection attribute. This was taken from www.GSMArena.com on September 26th 2013. The phone thickness attribute was missing from both these sources. It was taken from comapareindia.in.com on November 5th 2013.
For the Micromax Canvas 4 A210, data was missing for the GPS attribute. This was taken from www.GSMArena.com on September 26th 2013.
For the Lava Iris 504Q, data was missing for the GPS attribute. It was taken from comapareindia.in.com on November 5th 2013.
Rights and permissions
Copyright information
© 2016 Springer International Publishing Switzerland
About this chapter
Cite this chapter
Truscott, P., Korns, M.F. (2016). Predicting Product Choice with Symbolic Regression and Classification. In: Riolo, R., Worzel, W., Kotanchek, M., Kordon, A. (eds) Genetic Programming Theory and Practice XIII. Genetic and Evolutionary Computation. Springer, Cham. https://doi.org/10.1007/978-3-319-34223-8_12
Download citation
DOI: https://doi.org/10.1007/978-3-319-34223-8_12
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-34221-4
Online ISBN: 978-3-319-34223-8
eBook Packages: Computer ScienceComputer Science (R0)