This page records a handful of Use Case Scenarios for the imagined data warehouse and client

Quick Look analysis

A vehicle trial has been conduced. The data has been retrieved from magnetic tapes and memory sticks. An data operator opens a GND import application and loads each data file. For each data-file I expect him to add:

• the name of the trial that was being conducted
• the data-format of that file

The importer will then load the data into the Warehouse, and the data operator will indicate to an analyst that the data is loaded.

The analyst will then browse to a web-page, and enter the name of the trial. The web-page will then show the route where the vehicle travelled, potentially overlaid on OpenStreetMap. In an adjacent browser frame the analyst will display a time-variable xy plot of the value for a temperature sensor during that same period.

transform data to standard format

We've received data as tab-appears data, but the columns aren't in our standard units, plus the longitude is expressed as 359.5678 instead of -0.4322. So we need to operate on columns of data. We may need to apply a scaling factor to one column, perform a units conversion on another (knots to m/s), and add -360 to the wonky longitude column.

Also, the user may view a time variable graph of the data, spot an erroneous observation, and either specify a new value, or interpolate/spline the value from other valid ones.

I picture this happening in a spreadsheet style grid, where the user selects a column and then specified the option to perform. The user is able to view the before/after effect of the operation in a graph/plot, and on completion save the new version.

The user may also wish to merge this fatal with another.

Deeper analysis via data export

Data flow into the warehouse as before.

In this scenario the analyst it going to use a legacy analysis tools to investigate the performance of a number sensors. He will open the GND export application, and select a block of data via:

• selecting a named trial
• selecting a vehicle id
• selecting a recording name
• selecting a time period or geographic area

He will then specify the export format required by his analysis tool, and the exporter will provide a compatible data file.

Subtle change to sensor format

The new version of sensor means that one attribute is recorded in metric instead of imperial units. A data operator inspects the new technical specification of the sensor. He identifies which attributes have changed, and agrees with analysts that production of a new import data-file format and warehouse data format is necessary. He then defines transformations that move the additional fields into the data warehouse, and export the data from the data warehouse. The export transformation may perform a metric to imperial conversion in order to support legacy analysis tool file formats.

Completely new sensor type is adopted

A new sensor is adopted, one which is nothing like the existing ones used. A data operator meets with analysts and agrees which of the observation attributes are to be stored in the warehouse, they may also agree on a down-sampling required for the sensor (whilst the sensor records at 100Hz only one observation per second). He then defines import and export transformations as above.