Eltabakh, Mohamed Y.
This investigation examined the value of using Apache MADlib for analytical operations versus developing functions and procedures for the same purpose using PL/SQL. Several functions were chosen arbitrarily from the MADlib repository. After implementation for each function was complete, performance testing was conducted to examine the accuracy and runtime of each PL/SQL function against its MADlib counterpart. Results showed using MADlib can be significantly advantageous for through data processing compared to direct implementation. Manually implementing analytical functions could be efficient for smaller queries however as the sample size increases MADlib handled the queries much more effectively.
Worcester Polytechnic Institute
Major Qualifying Project
All authors have granted to WPI a nonexclusive royalty-free license to distribute copies of the work, subject to other agreements. Copyright is held by the author or authors, with all rights reserved, unless otherwise noted.