Considering metonymy in analytics

As a metonym, an object stands for itself; it does not symbolize, or stand for, or call for similarities with another object- it exists as its own object without representation.  The realm of analytics, which inherently relies on numbers, is ripe with metonymy and the strength of this structure is actuality.  The realism described by analytics is truth, plain and simple.  However, this is a truth that is only as righteous as the algorithm, the harvest of data, and methodology on which the algorithm is constructed and presented.  The interpretation of the analytics and discriminatory course of action pursued following the presentation of the analytical data is subjective and wrought with the complexities of human prioritization.  Within the data, there exist arithmetic truths, but the deduction may not be predictable because the deduction is based on contextual and situational input that are not captured in the analytic function.  Here we realize a tango-esque dance of the objective and subjective with the realm of analytics; a realm that grows ever mustier and shadowy as it welcomes any ‘big data’ guests into the salon.

We know from quantum mechanics that the measurement of a single variable will perturb the findings of an entire experiment.  In the case of analytics, even the choice of variables for presentation immediately guides a harvest of data along the conveyor belt and nearly predetermines the visual impact of the output crop.  Depending on the relay context, the type of graphing method for relay should almost be chosen before the data set to limit the desired impact of the information.  For example, showing an increasing trend would demand a bar or line graph and most certainly not a pie chart.  The act of choosing the architecture for display of the culled data becomes as artistic a practice as choosing the data for display at all and the process of drawing and proof sketching what form interpretable data could take becomes necessary if the data is to meant to influence a particular audience.  It is only through this often lengthy iterative process of data grooming and revelation that a reality begins to settle, or at least approach the asymptote, as a true metonymic structure of the investigation; the inherent and inevitable conclusion nears a solid form in time past the specter of metaphorical guessing and inference.

We must consider an event-driven architecture in the interface of the data’s presentation and constantly carry scrutiny for fraud.  There is caution toward the cardinality of any collection of data, since there is never a complete data set or real closed field post any complex event processing or unveiling of large information sets.  At most we can hope for a distant service and semblance of metonymy- that the sphere of Jupiter will not relent casting a sense of awe with its moons ever lurking in shadow tidal-cast in blissful orbit.

The interfaces that will be most useful in the capture, analysis, and relay of big data will be those that are of definiteness of purpose and of clear design toward a focused set of users, interpreters, and champions of findings.