Sunil S. Ranka's Weblog

Superior Data Analytics is the antidote to Business Failure

Posts Tagged ‘Best Practices’

Machine Learning Algorithm Cheat Sheet

Posted by sranka on April 11, 2017

With so many algorithms around its always a struggle to find out which algorithm could be suitable for the problem statement, I want to solve. Microsoft has done an amazing job to start with. Please find attached  Machine Learning Algorithm Cheat Sheet .

Screen Shot 2017-04-10 at 7.23.12 PM

Hope This Helps

Sunil S Ranka


Posted in Big Data, sunil s ranka | Tagged: , , , , , | Leave a Comment »

OBIEE Answers (Ad-Hoc Reporting) Best Practices

Posted by sranka on January 7, 2010

Wishing you all a very Happy New Year 2010!!!! . I was waiting for some exciting topic to start with my 2010 blogging. Before holidays I had a deliverable of documenting best practices for Answers implementation (Ad-hoc reporting). Answers, a topic close to me. Since requirements keep changing as new ideas starts flowing in from the end users, OBIEE ANSWERS comes is key for success of any operational and tactical Business Intelligence project.

OBIEE ANSWERS Ad hoc reporting empowers power business users to create their own reports and perform in-depth analysis of real-time information right in their Web browsers. Sometimes, business users need reporting capabilities for one-time questions or want to build specialty reports without having to request that the IT department create a new report for them. Ad hoc reporting allows end users to get answers on demand and customize their queries so that their reports contain only the information needed.

Best Practices

  • Getting critical information from the right people at the right time, detail Business Requirements analysis is critical to the success of Ad hoc reporting.
  • Ad Hoc reporting user will have access to most granular level of data, hence data and Object level security is high priority item.
  • A single security model should be used for both canned and ad hoc reporting
  • Empower and educate key end users on content and OBIEE to develop their own reports in a managed Ad hoc report environment.
  • Don’t release all the detail measures in the first phase, have phased approach for releases.
  • A separate OBIEE sas and web servers are preferred for better performance and scalability.
  • Join Fact and Dimension using numeric key for better performance.
  • Don’t overload the user with too many Fact and Dimension, keep the Dimensional modeling simple with few facts and Dimension for ease of use.
  • Training for comprehensive understanding of the subject area that will be queried should be given to End Users.
  • Educate users on restricting number of rows displayed on browser to less than 25.For detail data use download to data option.
  • Measures and Dimension should be logically grouped for better user experience.
  • Performance and detail unitesting of each and every measure with all the dimensions.
  • Create and save variables and filters for returning the most relevant results.
  • Basic and advanced statistical calculations needs to be done at the repository level
  • Answers access should be restricted only to power users.
  • Keep track of Answers subject area using Usage Tracking for improvements and enhancements.
  • Keep an eye on the db so that users don’t use most of the resources bu using some naked queries.
  • In past we have seen that when users are trying to download huge data SAS server crashes, so please make sure you have enough storage space defined for — DATA_STORAGE_PATHS,SORT_MEMORY_SIZE,SORT_BUFFER_INCREMENT_SIZE,VIRTUAL_TABLE_PAGE_SIZE — variable in NQSConfig.INI confige

Hope This Helps


Sunil S Ranka
Blog ::
“Superior BI is the antidote to Business Failure”

Posted in ..., OBIEE | Tagged: , , , , , | 1 Comment »

%d bloggers like this: