We describe a novel method for In-Database Machine Learning (IDBML) We contribute a process for . SQL-codegeneration in Python using template macros in Jinja2 as well as the prototype . We measured accuracy and .computation time . Our multidimensional probability estimation was significantlymore accurate than Naive Bayes . However, our method was 2-3% less .accurate than the best current state-of-the-art methods we found (decisiontrees and random forests) and 2- 3 times slower for one in-memory dataset . Yet, this fact motivates for further research in accuracy improvement and in IDBML with big data and larger-than-memory datasets. Yet,this fact motivizes for further . data . The authors also suggest further research to improve the accuracy of our method and system using the . database’s ‘

Author(s) : Michael Kaufmann, Gabriel Stechschulte, Anna Huber

Links : PDF - Abstract

Code :

Keywords : accurate - method - database - accuracy - sql -

Leave a Reply

Your email address will not be published. Required fields are marked *