Hi!
I do machine learning and statistical analysis in R; and lately, my company has been asking for the deployment of some trained models for certain purposes. As it is very explicit with deploying applications with sensitive data; it restricts ourselves to Power BI as the go-to tool.
I had some ideas in mind and wanted to double check here about the feasibility of them. Please, do assume I train my machine learning models in R separately and I import it as an object to be deployed outright in an R script of any type, like one we can find on Power BI. If that is the case, could I do the following use cases?:
Analyzing a PDF that is input for a classification task, being either multiple or binary. It is important to assess feasibility of some ingestion mechanism of PDF's here that is custom... Is that feasible?
selecting a specific row within an existing dataset within the Power BI report in order to run a regression / classification prediction on it. Is it possible?
I am not very acquainted with Power BI as a tool for doing machine learning, but I realized in many ways; if these simple use cases are feasible, we could develop a tool bypassing some work that approaches like Shiny apps could entail. Is there any possibility on getting further assistance with this feasibility? I would really appreciate it!
Thanks!