Page 17 - newDATAmagazine | 04>08>2021
P. 17
newDATA
MAGAZINE
Big Data WORLDS
How big must a problem be to need a
Big Data Solution?
The Cosmos? With all its galaxies, stars and
planets? With all its complex and unreachable
physical phenomena, relying only in data to be
understood? And why is this so different from
what we have right here, at arm's length?
Human activity in general, the frantic,
diverse and multidimensional, with all the data
that comes from it, seems to be a good
candidate any day of the week.
Or the very small, like subatomic particles in
the LHC, at CERN, where there's massive
collection of information in a very tiny fraction
of time from a very complex and difficult to the most interesting development in
observe source? business in the last decade or so, like
Though we still tend to see Big Data as an Uber, AirBnB, but also others, where the
extreme use case, where the word “Big” seems possibility of collecting information from
to be more of a marketing obstacle than a a wide variety of sources, in a wide variety
marketing enabler, many are the situations of formats, has opened up a much more
where we can apply it, really. Not just extreme diverse and competitive market, paving
ones. It all comes down to granularity, the way for much more empowered
dimensionality and speed. Volume, actually, consumers and lightweight, but strong
comes last, since it will depend on the latter. competitors.
Some of the most common use cases are: Ÿ Customer Churn, something of an oldie
Ÿ Fraud detection on credit cards, mobile but goldie, like fraud detection. In general,
phones – the usual; behavior is a good candidate for this
solutions, since they are multidimensional
Ÿ Sentiment Analysis, even more so with problems, where you also need many data
text, image and video, possibly even points across time and context, which in
cross-referenced;
turn takes the problem to another level.
Ÿ Retail consumer goods, namely in large You usually don't know where the data you
commercial stores, B2C, as I'll explain need is, so you have to collect an overload
further along the article. of it, to find the right data points that will
Ÿ Service mediation, as it has been one of help you solve the puzzle.
newDATAmagazine.com 17