The Application of Interpolation Technique to Market Research

Question

Hi All,

I know the interpolation tools in ArcGIS are widely used for the estimation of weather condition and other environmental area.

However, I am wondering if it is appropriate to apply interpolation techniques to the area of market research. For example, I have some survey data of individuals' willingness to purchase a certain product in the given study area. After I geocode those individuals, I wonder if I can create a prediction map to predict the location of the potential customers in the study area.

I understand you have to meet certain assumptions such as auto-correlation and normal distribution to conduct interpolation. But let's assume my data are pretty robust, any thoughts on the appropriateness of using interpolation in this case?

As always, your inputs are much appreciated.

Thanks,
Jason

MarcoBoeringa · Answer

Thanks for the feedback!
That's good to know and I'll explore more into this.

Jason

Hi Jason,

Although Eric is right that from a purely practical point of view "if the data meets the assumptions ... kriging should work", I would really like to emphasize the small caveat Eric placed in between these quotes and that I hereby rephrase:

"Does it make sense to interpolate your particular dataset???"

People all to easily grab to the first available tool that seems suited, while there may be better, and from a scientific point of view sounder, methods to accomplish what you want, e.g. create a map of potential customers.

Interpolation always seems attractive when having a point dataset, as it is an "easy" way to get a surface, but is it sound from a logical and scientific point of view?

I can't answer that question for you, but you might be... based on your knowledge and questions like the one below

E.g.  do you find it likely that two people nearby have a higher likelihood to buy a certain product than two people further apart (spatial autocorrelation)? Does this almost always hold? (seems unlikely with market research data, two neighbours may have a completely different social background and status, but for good continuous environmental datasets it usually does)

My gut feeling says that this type of "social" data is not the best suited for interpolation, and that you should be careful in your decision to use it and interpret results.

I really shouldn't ignore general statistical analysis (multi-variate) like can be done with general statistics packages like NCSS and SPSS, and than use any found statistical relationships to classify other datasets with your market data to answer your question of "where potential customers may be".

Yes, it is more laborious than the "one-stop point-to-surface option" that interpolation seems to offer, but it might be more sound from a scientific and statistical point of view.

That is all not to say it can't be done, just that you should be careful in your decision and not ignore other options. And ESRI has of course "Business Analyst". Maybe time to look at that?

Marco

LokKwan · Answer

Thanks for the feedback!
That's good to know and I'll explore more into this.

Jason

WillyG_0-1780865017001.png

image.png

Eric6346 · Answer

Whether kriging will work really depends on your data and the question you're trying to answer.  We generally use environmental data in examples because this kind of data most often meets the kriging assumptions, but there is really no restriction on the source of the data.  If the data meets the kriging assumptions and it makes sense to interpolate, kriging should work.

Zurich_2.3.jpg

NolanNiedermeyer_0-1780862303385.png