# What method should I use for prediction of NUMBER_OF_ORDERS(TIME, LAT, LNG)?

I'm running a pizza service and would like to predict a number of orders for every hour interval during a day per location (basically where we should deliver the pizza) in future.

And I've got a pretty big data set of past orders: pairs of [timestamp, location], where location is a pair of latitude and longitude.

So I've got NUMBER_OF_ORDERS(TIME, LATITUDE, LONGITUDE) function. What's the most simple yet accurate method to use for this task?

I was told that linear regression is a bad choice because number of pizza orders is not a linear function of either time or locations.

• Spatio-temporal data analysis is its own sub-field. This textbook is a good place to start. amazon.com/Statistics-Spatio-Temporal-Data-Noel-Cressie/dp/…
– Sycorax
May 16, 2018 at 15:53
• This is a hard problem to solve, and it depends on how much accuracy you want. For a pizza service a linear regression model might work well.
– knk
May 16, 2018 at 15:53
• @Sycorax could you recommend some nice repo on github / some concise research paper? I've found a huge number of formulas and they're all pretty complicated (+ without the code itself). I even tried to search for 'spatio-temporal data bike nyc dataset' (append some random dataset trying to find a small pet project with code). May 16, 2018 at 18:41
• @knk do you really think LR model will work? E.g. if I fix location and select timestamps within one day I probably should get (/\ instead of a straight line, where '/\' means non-zero number of orders in the afternoon and about zero in the morning). May 16, 2018 at 18:43
• @J.Doe This is not a field which has yet coalesced around simple APIs in the same way that sklearn collects common, generic classifiers. The reason is that there are complex relationships among observations which are explicitly a part of the modeling problem. That makes the math hard, and it makes designing generic software hard, too - because you need to present those relationships and model them.
– Sycorax
May 16, 2018 at 19:31