<s>
Data	B-Algorithm
thinking	I-Algorithm
is	O
a	O
product	O
design	O
framework	O
with	O
a	O
particular	O
emphasis	O
on	O
data	O
science	O
.	O
</s>
<s>
In	O
the	O
context	O
of	O
product	O
development	O
,	O
data	B-Algorithm
thinking	I-Algorithm
is	O
a	O
framework	O
to	O
explore	O
,	O
design	O
,	O
develop	O
and	O
validate	O
data-driven	O
solutions	O
.	O
</s>
<s>
Data	B-Algorithm
thinking	I-Algorithm
combines	O
data	O
science	O
with	O
design	O
thinking	O
and	O
therefore	O
,	O
the	O
focus	O
of	O
this	O
approach	O
includes	O
user	O
experience	O
as	O
well	O
as	O
data	B-General_Concept
analytics	I-General_Concept
and	O
data	O
collection	O
.	O
</s>
<s>
The	O
term	O
was	O
created	O
by	O
Mario	O
Faria	O
and	O
Rogerio	O
Panigassi	O
in	O
2013	O
when	O
in	O
a	O
book	O
about	O
data	O
science	O
,	O
data	B-General_Concept
analytics	I-General_Concept
,	O
data	B-General_Concept
management	I-General_Concept
,	O
and	O
how	O
data	O
practitioners	O
were	O
able	O
to	O
achieve	O
their	O
goals	O
.	O
</s>
<s>
Data	B-Algorithm
thinking	I-Algorithm
is	O
the	O
understanding	O
that	O
a	O
solution	O
to	O
a	O
real-life	O
problem	O
should	O
not	O
be	O
based	O
only	O
on	O
data	O
and	O
algorithms	O
,	O
but	O
also	O
on	O
the	O
domain	O
knowledge-driven	O
rules	O
that	O
govern	O
them	O
.	O
</s>
<s>
Data	B-Algorithm
thinking	I-Algorithm
asks	O
whether	O
the	O
data	O
offer	O
a	O
good	O
representation	O
of	O
the	O
real-life	O
situation	O
.	O
</s>
<s>
Data	B-Algorithm
thinking	I-Algorithm
is	O
the	O
understanding	O
that	O
data	O
are	O
not	O
just	O
numbers	O
to	O
be	O
stored	O
in	O
an	O
adequate	O
data	B-General_Concept
structure	I-General_Concept
,	O
but	O
that	O
these	O
numbers	O
have	O
a	O
meaning	O
that	O
derives	O
from	O
the	O
domain	O
knowledge	O
.	O
</s>
<s>
Data	B-Algorithm
thinking	I-Algorithm
is	O
understanding	O
that	O
any	O
process	O
or	O
calculation	O
performed	O
on	O
the	O
data	O
should	O
preserve	O
the	O
meaning	O
of	O
the	O
relevant	O
knowledge	O
domain	O
.	O
</s>
<s>
Data	B-Algorithm
thinking	I-Algorithm
analyzes	O
the	O
data	O
not	O
only	O
logically	O
but	O
also	O
statistically	O
,	O
using	O
visualizations	B-Application
and	O
statistical	O
methods	O
to	O
find	O
patterns	O
as	O
well	O
as	O
irregular	O
phenomena	O
.	O
</s>
<s>
Data	B-Algorithm
thinking	I-Algorithm
is	O
understanding	O
that	O
problem	O
abstraction	O
is	O
domain-depended	O
,	O
and	O
generalization	O
is	O
subject	O
to	O
biases	O
and	O
variance	O
in	O
the	O
data	O
.	O
</s>
<s>
Data	B-Algorithm
thinking	I-Algorithm
is	O
understanding	O
that	O
lab	O
testing	O
is	O
not	O
enough	O
,	O
and	O
that	O
real-life	O
implementation	O
will	O
always	O
encounter	O
unexpected	O
data	O
and	O
situations	O
,	O
and	O
so	O
improving	O
the	O
models	O
and	O
the	O
solution	O
for	O
a	O
given	O
problem	O
is	O
a	O
continuous	O
process	O
that	O
includes	O
,	O
among	O
other	O
activities	O
,	O
constant	O
and	O
iterative	O
monitoring	O
and	O
data	O
collection	O
.	O
</s>
<s>
Even	O
though	O
no	O
standardized	O
process	O
for	O
data	B-Algorithm
thinking	I-Algorithm
yet	O
exists	O
,	O
the	O
major	O
phases	O
of	O
the	O
process	O
are	O
similar	O
in	O
many	O
publications	O
and	O
could	O
be	O
summarized	O
as	O
follows	O
:	O
</s>
<s>
Design	O
thinking	O
principles	O
in	O
the	O
context	O
of	O
data	B-Algorithm
thinking	I-Algorithm
can	O
be	O
interpreted	O
as	O
follows	O
:	O
when	O
developing	O
data-driven	O
ideas	O
,	O
it	O
is	O
crucial	O
to	O
consider	O
the	O
intersection	O
of	O
technical	O
feasibility	O
,	O
business	O
impact	O
,	O
and	O
data	O
availability	O
.	O
</s>
<s>
To	O
scope	O
data	O
and	O
the	O
technological	O
foundation	O
of	O
the	O
solution	O
,	O
practices	O
from	O
cross-industry	O
standard	O
processes	O
for	O
data	O
mining	O
(	O
CRISP-DM	B-Algorithm
)	O
are	O
typically	O
used	O
at	O
this	O
stage	O
.	O
</s>
<s>
CRISP-DM	B-Algorithm
)	O
.	O
</s>
<s>
Solution	O
feasibility	O
and	O
profitability	O
are	O
proven	O
during	O
the	O
data	B-Algorithm
thinking	I-Algorithm
process	O
.	O
</s>
