Hands-On Industrial Internet of Things

Step 4 – packaging and deploying

We can now develop a unit test and a regression test to check the quality of our analytics. The following algorithm develops a simple unit test that checks the anomaly detection procedure that we have developed with a predefined dataset:

import unittestfrom my_anomaly_detection import search_anomaliesclass MyTest(unittest.TestCase):    def test(self):    d=np.array([10, 10, 10, 10, 30, 20, 10, 10])    a=search_anomalies(d,2)    a=a['anomalies']    self.assertListEqual(a[1], [4,30])if __name__ == '__main__':    unittest.main()

To deploy the analytics on the cloud, we have to decide whether to use a cold-path or a hot-path. The developed algorithm doesn't need a lot of data and should run on the data that we have available. We ...