Machine learning tools and frameworks
Machine learning adoption is rapidly increasing among technology and business organizations. Every organization is actively strategizing on how to capitalize on their data and use it to augment their client's experiences and build new businesses. When it comes to tools or frameworks for Machine learning, there are many open source and commercial options on the market. The new age tools are all built to support big data, distributed storage, and parallel processing. In the next chapter, we will cover some aspects of handling large scale data in the context of Machine learning.
At a very high level, there are three generations of Machine learning tools.
The first generation of Machine learning tools is focused on providing a richness of the Machine learning algorithms and supporting deep analytics. These tools haven't been built to focus on handling large scale data or for supporting distributed storage and parallel processing. Some of them still handle volumes as a result of their support for vertical scalability. Some of the tools that come under this category are SAS, SPSS, Weka, R, and more. Having said that, most of these tools are now being upgraded to support big data requirements too.
The second generation tools are focused on supporting big data requirements, most of them work on the Hadoop platform, and they provide capabilities to run machine learning algorithms in a MapReduce paradigm. Some of the tools that are categorized here are Mahout, RapidMiner, Pentaho, and MADlib. Some of these tools do not support all the machine learning algorithms.
The third generations tools are the smart kids on the road, breaking the traditional norms of operating in batch mode, supporting real-time analytics, providing support for advanced data types of big data, and at the same time supporting deeper analytics. Some of the tools that are categorized under this are Spark, HaLoop, and Pregel.
In Chapter 4, Machine Learning Tools, Libraries, and Frameworks, we will cover some of the key machine learning tools and demonstrate how they can be used based on the problem's context. Implementation details for tools such as R, Julia, Python, Mahout, and Spark will be covered in depth. Required technology primers and installation or setup-related guidance will be provided.
