Data Screening

What is abnormal data?

Abnormal data refers to values measured at a time when air quality measuring equipment was not working correctly (e.g., malfunction or under calibration).

Why to filter out abnormal data?

When a set of measurement equipment is not working correctly, the air quality readings may be excessively high or low so that the public may become skeptical about the data as a whole. Therefore, abnormal data should be identified and sorted out to make sure that only normal data should be provided to the public, thereby enhancing the credibility of the data.

How to sort out abnormal data?

Abnormal data is automatically sorted out in the following two steps through computer programming:

  • Step 1 : Determine whether the data is abnormal, only when the information on the conditions of measurement equipment (e.g. calibration, inspection, or malfunction) is available.
  • Step 2 : Regard the collected data as abnormal one when it exceeds the normal range or rate of change.
Fault in connection with measurement equipment -> Malfunction -> Under calibration -> Readings exceeding the normal range of measurement -> Data acquisition rate of less than 75% -> Excessive fluctuation of data -> Readings excceding the permittable range of pollution -> Consistently equal values -> Normal data