Native scoring is a much overlooked feature in SQL Server 2017 (available only under Windows and only on-prem), that provides scoring and predicting in pre-build and stored machine learning models in near real-time. Depending on the definition of real-time, and what does it mean for your line of business, I will not go into the definition of real-time, but for sure, we can say scoring 10.000 rows in a second from a mediocre client computer (similar to mine) . Native scoring in SQL Server 2017 comes with couple of limitations, but also with a lot of benefits. Overall, if you are looking for a faster predictions in your enterprise and would love to have a faster code and solution deployment, especially integration with other applications or building API in your ecosystem, native scoring with PREDICT function will surely be advantage to you. Although not all of the predictions/scores are supported, majority of predictions can be done using regression models or decision trees models (it is estimated that both type (with derivatives of regression models and ensemble methods) of algorithms are used in 85% of the predictive analytics).
Jun-6-2018, 06:36:35 GMT