Those of us who were data science practitioners before the ubiquity of buzz words like “Fourth Industrial Revolution”, remember a time when data science was being described as “utter hubris” by some of the worlds leading quants. All these buzz words, coupled with misplaced applications of data science have turned it into a fairly nebulous concept. Which is a problem.
In facilitating my data science workshops I've realized there are two main groups of audiences and learners. Those who see it as pure wizardry and those who see it as an inferior derivative of statistics which has been sullied by computer science and sci-fi films. Both these groups are mistaken in their assumptions but neither are completely unjustified in making them.
Perhaps as proponents of this sometimes fabled fourth industrial revolution need to work towards bringing data science to the fingertips of the average Joe who just wants to use AI to do some cool stuff. Maybe the only way to de-mystify data science is to cut away the complex code and the iterative processes.
Which brings me to Intren, a front-end interface environment for R Packages.
Intren's base algorithm reads package documentation and gives the user an organised representation of the packages' functionality. This includes extracting the names of each function in the package; as well as their respective descriptions, arguments and syntax. Intren then allows the user to execute these functions using an easy-to-use interface without writing any code.
Below is basic demo of the rpart package being used by Intren to fit a decision tree to predict employee KPI scores. Any package on the CRAN repository with documentation can be used by Intren without writing any R code. I recently presented this demo at a machine learning workshop at EY and for the first time in a workshop people gained an intuitive understanding of ML in 5 mins.
Decision Tree Demo from Mkhuseli Mthukwane
Would love to know some of the related projects some of you are working on!