Analytics

Ever wondered how Google comes up with exact suggestions for your search in Google Maps, irrespective of you changing the location. It employs predictive analytics based on past searches, what is popular now, or your location are at that point of time.

 Analytics is about analysing the data by slicing and dicing it and representing as per the business need. Next generation Analytics is for generating “actionable insights (e.g. Machine Learning)” from data sources - be it “historical data” or “real time” data.

Analytics is the necessity of everyone. The same sort of analytics is not needed by/for every organization. Below is a matrix that defines analytics maturity. Based on the current stage of the analytics maturity index, the next course of actions can be taken to take the maturity index to next level.

The Analytics ecosystem consists of multiple components,

1. Data transformation

Data transformation has evolved due to evolution of data from structured data to unstructured data,

1.a  Traditional ETL approach of transformation of data

§ ETL stands for “extract”, “transform” and “load”. This is for traditional systems where data was to be converted from one source system to other, where the output schema was structured and well-defined.

1.b  New ELT approach in realm of real-time analytics

§ ELT stands for “Extract”, “load” and “transform” more called as “schema on read”. The data is loaded from data source to a data store which often termed as a “data lake”. Data lake necessarily contains unstructured data. In this approach, the target schema is defined as per use case and the transformation happens at the last leg. The data in “data lake” is transformed as per the requirement of application fetching the data, e.g. data document is created as per specifications needed by the end application.

2. Visualization –

The visualization layer depicts what user wants to decipher out of the available data. It can be bespoke or real-time, can be created by Microsoft as well as open-source tools, few below,

o  Tableau – Tool to connect with any data source and create reports with drag and drop.

o  Qlik – Guided analytics tool

o  Logi Analytics – Completely customizable JavaScript based tool

o  MSBI-SSRS – SQL Server Reporting Services is Microsoft tool for creating reports

3. Types of Analytics

Below are the types of analytics

3.a Descriptive analytics

§ Descriptive analytics is initial stage of data processing to create a summary of available data to gain useful insights and prepare the data for further analysis

3.b Prescriptive analytics

§ Prescriptive analytics is about deciding the next best action based on available data. It can be related to predictive as well as descriptive analytics

3.c Predictive analytics

§ Predictive analytics is about making predictions using data based on techniques like data modelling, statistical analysis, machine learning, artificial intelligence etc.

Additionally,

3.d Edge analytics

§ Edge analytics – as the name suggests is done before the data is stored so that the latency of data storage and retrieval is reduced. This is mostly used for IoT (Internet of Things) where data comes from devices. Edge analytics is supported by any cloud which supports IoT.

Some food for thought !!!

Today, personal data of individuals is available in parts to multiple entities like Google, Facebook, Linked-in, Twitter, Amazon, Apple etc. They use the data for their specific business needs. What if tomorrow they all form a consortium and share the data across (with user permissions) to create whole from parts. There can be lot of interesting insights generated out of this humongous data…

This is one of the reasons why data is rightly called the next oil…