Getting Anomalylikelihood from Multi-field anomaly analysis

Engineering NuPIC
1

Hi, Forum!

I’m doing a project of detecting anomalies from data composed with more than 2 sensor fields.

Modified_Time		Sensor1	Sensor2
datetime	string	string
T			
2016-01-01 0:00	A	K
2016-01-01 0:30	B	Y

In using anomalyLikelihood.anomalyProbability function,
what should I put for the “actualValue” parameter in the following code?
Actual value of sensor 1 or sensor 2? (By the way, as shown above, the sensor values are categorical (string) values.)

from nupic.algorithms import anomaly_likelihood
anomalyLikelihood = anomaly_likelihood.AnomalyLikelihood()
likelihood = anomalyLikelihood.anomalyProbability(actualValue, anomalyScore, timestamp)```

Please help!!
Thank you in advance.

2

I think either will work.

3

Oh, thank you very much!!