Subscribe to DSC Newsletter

We have all heard about the case where Target knew about a teenager being pregnant before her father knew.  I do not know the calculation of the formula for producing the results.  However, I question if this is truly predictive analysis.  In my opinion this is just an assumption, hypothesis or a guess based on the items that were purchased by the teenager.   She was not predicted to be pregnant because she was already pregnant. For example, if I purchased ice cream, potato chips, cookies, and bon bons every week, based on my purchases, would you assume that I am over weight or in poor health?    A predictive assessment would be that if I continue my weekly purchasing pattern, I have the potential of becoming a diabetic in 5 years. 

So, my question is; are we using assumptions as predictions?

Views: 2855

Reply to This

Replies to This Discussion

The article link given by Kuber Chaurasiya , from another group, gave me the answer I was looking for.

 I am still not so convinced that it is a prediction; but rather she was selected because her information is listed in Target’s baby registry at the store.  Probably all women that are listed in the registry are receiving information on sales of baby items.

The article states:

“Target assigns every customer a Guest ID number, tied to their credit card, name, or email address that becomes a bucket that stores a history of everything they’ve bought and any demographic information Target has collected from them or bought from other sources. Using that, Pole looked at historical buying data for all the ladies who had signed up for Target baby registries in the past. “


It is a prediction if the model was trained on a dataset identifying women who were actually pregnant and was cross validated as well

It is an assumption if a Marketing expert thought of few indicators that would predict pregnancy but could not test it out (or used Baby registry, as mentioned, which in itself is an assumption, albeit much more practical)

Predictions are hypothesis that have been validated. Although in a real sense, assuming that the prediction will hold in future, regardless of how good the model is, is still an assumption

It may be a prediction.  However, I think it is over-kill and a little unnecessary, maybe in a harmful way.  

Thanks for mentioning me Dorothy :)

Dear Dorothy,
Good evening. I read your post it is really good question. In my opinion, we're discussing about prediction, forecasting and other statistical technique. But, every thing depend on past data. If your past data give indication about those parameters then you should consider then, but at the same time you should be aware field knowledge which will help to avoid having baise data. In above case we should rethink about the parameters as well as mix of the data i.e. The ratio of preganent women and that of Not pregnant has to be equal.


On Data Science Central

© 2021   TechTarget, Inc.   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service