Data Outliers in NI IoT Sensors

[zVin]

Hi all,

Good morning,

1) In general, may I know why are there many outlier data when using IoT sensors ?

You know the outliers can be so many, that they form data clusters. You can read more on https://www.cio.com/article/3151081/internet-of-things/worm-on-the-sensor-what-happens-when-iot-data...

2) What is so special on IoT sensor that it produce so many outliers compared with traditional sensors ?

3) Does it have to do with the IoT sensor technology or it is a quality issue ? 

Hope to hear from you. Thank you very much.

Regards, 

Vincent Lee

[Albert.Geven]

For me it is standard quality. On instruments that measure you always have to be critical. Most of them are so good that people forget about calibration and checks for years. But calibration also detects slow changing errors and will pay itself back.

On the other hand. It is costly and sometimes measuring with more instruments and comparing is cheaper especially in iot.

[altenbach]

• Please don't post the same message in several locations. Once is enough.
• Please clean up your links and don't paste links directly from your e-mail.
• Please don't link to deceptive sites that serve tons of advertisements with no real content. This is basically a blatant ad mill, IMHO. Nobody should click on it.
• Are you actually involved with IoT and have problems with outliers? In that case explain your setup and problems you experience. Are you even using NI hardware or software? No links to questionable sites needed.

[zVin]

Albert,

Thank you for the reply on IoT sensor.

May I know are there any standards to ensure quality of IoT sensors ?

When measuring temperature with IoT sensor, does ASTM standard E2846  https://www.astm.org/Standards/E2846.htm ensure quality of IoT sensor to have little outliers ?

Please reply. Thank you.

Vincent

[Albert.Geven]

Outliers are not only due to the sensor but also to the environment.

Or even small changes in positioning of a sensor close to a heated region e.g. in an oven you need to measure at more than 1 point to see the effect of the position in the back or close to the front door.

So outliers are always possible, just be sure you calibrate a system, not just a sensor.