Big Data and Business Analytics

171

11

Extracting Useful Information from

Multivariate Temporal Data

Artur Dubrawski

INTRODUCTION

Many of the important big data sets encountered in practice assume the

form of a record of transactions. Each entry in such data typically includes

date and time of an event (such as a nonprescription drug purchase trans-

action or a record of a repair of a vehicle) and a potentially large number

of descriptors characterizing the event (e.g., the type, dose, and quantity of

the medicine sold or the model year, make, conguration, and description

of failure of the vehicle). Time- stamped transactional data can be used

to answer various questions of p