User`s guide
R2014b
2-34
Performance and Big Data
Big data analysis on your desktop that can scale to Hadoop with
mapreduce
The mapreduce function enables analysis of data sets that do not fit in your computer’s
memory. It is used to process large data sets on your desktop, and can also be extended
to run on Hadoop
®
to process big data. MapReduce is a powerful technique for applying
data processing methods to very large data sets, from simple statistics to complex
machine learning algorithms.
For more information, including a selection of examples, see MapReduce.
The functionality of mapreduce extends beyond MATLAB with the following products:
• Access relational databases using Database Toolbox™
• Increased performance on desktops with Parallel Computing Toolbox
• Scaling up to Hadoop using MATLAB Distributed Computing Server™
• Create deployable archives or standalone applications that run against Hadoop using
MATLAB Compiler™
Improved performance for sorting categorical data with sort
The performance of the sort function improves for large categorical array inputs.
typecast function performance improvements with long vectors
The performance of the typecast function improves for long input vectors.