Difference between revisions of "Data Scaling"

From Open Risk Manual
 
Line 3: Line 3:
 
set so that it conforms with a set of requirements.
 
set so that it conforms with a set of requirements.
  
Data Scaling will typically be required when the [[Measurement]] process generating the data sets is using a temporal, spatial or other scope boundary that is not aligned with the usage requirements (e.g., different time interval). The scaling adjustment, while a mathematically simple operation may imply [[Model Assumptions]] that may or may not be true.
+
Data Scaling will typically be required when the [[Measurement]] process generating the data sets is using a temporal, spatial or other scope boundary that is not aligned with the usage requirements (e.g., different time interval).  
 +
 
 +
== Issues and Challenges ==
 +
The scaling adjustment, while a mathematically simple operation may imply [[Model Assumptions]] that may or may not be true. Data derived from scaling should ideally be cross-checked with historically available local data or data derived from other means of estimation such as extrapolation or interpolation based on  historical local data, or modelling.
  
 
== See Also ==
 
== See Also ==

Latest revision as of 10:44, 5 January 2022

Definition

Data Scaling is a loose term that refers to data transformation activities that aim to improve the informational content of the data by adjusting an existing data set so that it conforms with a set of requirements.

Data Scaling will typically be required when the Measurement process generating the data sets is using a temporal, spatial or other scope boundary that is not aligned with the usage requirements (e.g., different time interval).

Issues and Challenges

The scaling adjustment, while a mathematically simple operation may imply Model Assumptions that may or may not be true. Data derived from scaling should ideally be cross-checked with historically available local data or data derived from other means of estimation such as extrapolation or interpolation based on historical local data, or modelling.

See Also