QUICK FACTS

HEAD OFFICE

Amrita Software & Technology Labs, IoT & Big Data research
Kollam, Kerala 690525
India

INDUSTRIES

Any Industry

PRICING

Free Trial

None

TYPE OF CUSTOMERS

For profit

SIZE OF CUSTOMERS

Small

 
Founders:
Shiju Sathyadevan
 
Year founded:03/04/2013
 
#Employees:40
Patents:1
Vendor size:medium
 

Overview

Masses of data is accumulating everywhere in different forms. Some in a very structured manner where as some others are in totally unstructured form. BI process are now challenged with the new trend of accommodating unstructured data as part of analytics data set. This includes customer feedbacks/comments, social media data etc. With IoT gaining its pace, streaming data from billions of sensors are going to flood the internet space. Applications are now posed with a new challenge of handling this highly dynamic data. Data need to be processed on the fly which may involve disk writes and/or in-memory processing. The need for crunching this large volume static and streaming data has resulted in new research that have produced new technologies to process data in parallel.

There are several processing frameworks out there. Hadoop ecosystem that caters for large volume static data, Apache Spark for in-memory analytics etc. In order to implement complex algorithms in Hadoop one need to write the code in Map Reduce. Also in order to fully leverage the in-memory power of Apache Spark developers need to build code using SCALA. Because of such restrictions Data scientists are very much depended on developers to prepare the data and interpret the data sets. This is usually a time consuming process. Whole purpose of building ABDF (Amrita Big Data Framework) is to reduce this programmer dependency and development time to Analytics.

Less

THE 3 KEY BENEFITS

  • 01 ABDF is a Big Data framework that brings the power of Hadoop and Spark under one roof. It supports a number of data a...
  • 02 Several computation modules like expressions (for applying custom computation onto an incoming row), MatchBox (can dy...
  • 03 Complex mathematical computations can be done by using the built in computation pool that can perform Matrix operatio...

THE 3 KEY BENEFITS

  • 01 ABDF is a Big Data framework that brings the power of Hadoop and Spark under one roof. It supports a number of data adaptors to fetch data from several different data sources. All algorithms are written from scratch as fully customizable bare bone algorithms. All pre and post processors that are needed to prepare the data before and after processing through machine learning algorithms are included as separate reusable components so that it can be used across multiple algorithms.
  • 02 Several computation modules like expressions (for applying custom computation onto an incoming row), MatchBox (can dynamically load data from different data sources for matching against an incoming data source) etc are inbuilt into the tool. Users can build models and share the models so that it can be used for predictions against new data sets.
  • 03 Complex mathematical computations can be done by using the built in computation pool that can perform Matrix operations (Add, multiply, transpose, inverse), vector operations like Dot Product, distance measures using Euclidean, Squared Euclidean, Manhattan, Chebyshev etc. Support for streaming data is provided. Message Queues like MQTT, ZMQ, Kafka can be used to publish and subscribe to streaming data with topic based authentication which can be processed over a time window.
Less




In order to contact Amrita Software & Technology Labs, you have to be a registered Datafloq user

Register here or Login








REQUEST SUCCESSFULLY SENT!