100%(1)1 out of 1 people found this document helpful
This preview shows page 3 out of 3 pages.
# Structurised data in form of R vectors and listsjsonRespParsed<-content(result,as="parsed") jsonRespParsed5.What are the different methods to deploy a model into production system? Ans:Before deploying any models into productions, we should consider following points:a.Programming language like Java,C, VB.b.Database and SQL scripts.c.Predictive model mark-up language.d.Schedulers. I. What are my requirements? (i.e. how many requests are you expecting per second, what latency is required…etc)II. How will I evaluate the model’s performance in production? (and how will I collect and store theadditional data from interactions) III. How frequently do I plan on re-training my model?IV. What are the data pre-processing needs? V. Will the format of the production input data differ drastically from the model training data? Willitcome in batches or as a stream? VI. Does the model need to be able to run offline? *Broadly the methods can be categorized into two:I. Online deployment models: Data mining tools like RevoDeployR or Orange or using Webframework.II. Offline deployment Modes: