How to start applying R in Enterprise. Practical example

    Publication of the presentation at the section of R meetup @ Moscow Data Science Major (Spring 2019) .
    The whole presentation is in pdf format .

    Why is this question relevant?

    Business cases are different, the technical essence is the same

    • Call Center Performance Analytics
    • Sales analytics, including forecasts
    • Antifraud system
    • Business process mining
    • Various audits (technical, financial)
    • Warehousing and logistics tasks
    • Activity-based costing
    • Business process monitoring
    • Log-based analytics
    • Capacity management
    • Text analytics (e-mail, service-desk)
    • Flexible dashboards and reports
    • "smart tires" between accounting systems (1C, ACS, SAP, ...) and executive
    • ...

    It is a continuation of previous publications .

    Practical observations

    • a lot of such tasks come down to mathematical manipulation of data (CRUD systems are beyond the scope, we consider precisely various processing and transformation);
    • 80% of data manipulation tasks can be quickly and efficiently solved "turnkey" by using the R tools;
    • in business, as a rule, tasks and requirements are quickly adjusted, incl. due to external factors or intermediate results obtained;
    • "modular" technologies take root well in IT; the construction of the "monolith" may take 2-3 years, which is comparable to the life of a small solution. It is much more efficient to quickly assemble a “modular” design, accumulate practical experience, and after 2-3 years build a new solution taking into account the knowledge gained and past changes in IT and business.

    Typical “urban legends” about R

    • R slow
    • R hard to read
    • R is for stat. calculations by complex algorithms
    • R is designed for interactive use.

    All this arises from a superficial study of the topic and the tools used.

    City legends - misconceptions from the 90s

    • R is a complete programming language, not a console calculator.
    • R acts well as a universal “glue” between various platforms and C components - it counts quickly!
    • The readability of the code depends on the experience of the developer. The modern style of R is metaprogramming. The code is compact and fast.
    • R is an ecosystem that allows you to implement a complete data processing cycle from importing data to providing AWPs and preparing presentations.

    Previous publication - "Using the computing power of R to test the hypothesis of equality of means . "

    Also popular now: