The article explains it's a mix of human observers and sensor data plus a lot of trial and error that has been done in order to get accurate prediction and to build trust.
What are the strategies to avoid or reduce false positives. I suppose the human observers are vetted? And the article suggests that there must be sufficient amount of sensor data for the likelihood to be big enough. However it fails to mention how many sensors and therefor data-points are collected.
Are there good resources on crowdsourcing data, ensuring certain quality of data, especially when it comes to limited data inputs?
What are the strategies to avoid or reduce false positives. I suppose the human observers are vetted? And the article suggests that there must be sufficient amount of sensor data for the likelihood to be big enough. However it fails to mention how many sensors and therefor data-points are collected.
Are there good resources on crowdsourcing data, ensuring certain quality of data, especially when it comes to limited data inputs?