Data Usage

Data is normally used in three ways:

• Operations
• Administrative
• Analytic

Operations

Operational data is needed to 'run the business': This is how a lab technician knows what they need to do next for the day (what is in their work queue). Automated systems replace the technician with an application, where a thermostat watches temperature events to determine if heating or cooling need to be adjusted. If the operations go correctly, the operational data has little long term value; other than confirming that plans were followed correctly. Besides using operational data, operations are often the source creating data as users record the start or completion of a workflow. Operations uses individual records and is closely tied to the data information and status (what is finished).

Administration

Administration uses data to track efficiently. Are we increasing or decreasing the number of widgets produced per day? How long did a process take, who is the most productive individual (or team), what is the most common cause of germination failure, is the failure rate of a sensor increasing? Administrative reporting is usually reported over a period of time, summarizing data by days, weeks or years. Administrative reports often compare one business cycle with another. These reports are used for decision making, making changes to business processes. Administrative questions are often fixed and asked over and over again. These are standard reports run on a frequent cycle, or dashboard charts for quick reference. There is little that is novel or creative. Administrative reports summarize records by date, and often focus on failure reasons (Status Qualifier). A greenhouse that may have hundreds of temperature sensors may have reports on mean-time between failures (which brand of sensor fails most frequently?, which locations are the most susceptible to failure?) and tie operational data into maintenance and inventory data.

Analytics

Analytic reporting is research and discovery. These are the ad-hoc complex questions that put the data to the test. These are usually novel queries that are looking for patterns or correlations. What is the relationship between stream height and trubidity? How fast does a crest move down stream, and what is the impact on turbidity? Operational workflows are of little value, with the focus on observations recorded as part of a workflow. Particulars of how the data was captured (what sensor used) is irrelevant, unless there are concerns about data quality. Records with a status other than 'Success' are ignored or discarded. Data is frequently combined with other sources such as weather data, topography, farming activity or construction projects. Location is frequently expanded with further detail. Latitude and Longitude may be the starting point for identifying watershed location (HUC numbers), which may be further expanded to stream segment (mile marker).