Sunday, May 17, 2015

Business Analytics @IIT


Yday, I attended 'Analytics for Business' session by Aaum Analytics at Indian Institute of Technology Park.  Itz pretty interactive and deep technical discussions.  In spite of weekend, significant set of participants came for the session, which reflects the thirst on Big Data technology.

Few take away points:

Recommended solution
  1. Persist raw inbound data in the original file format with cheap cost, if required
  2. Process the inflow data; summarize those information; then persist in traditional DB
  3. It helps the reporting layer to extract & present the report in better way

Challenges to Adapt
  1. Return on Investment (RoI) business justification
  2. Key Performance Indicator (KPI) missing accuracy
  3. Unavailable Talent pool
  4. Customer, Analyst, Developer growing requirements gap

Assessment Attributes for Analytic
  1. Demographic
  2. Behavioral
  3. Psycho-graphic
  4. Processing parameters

Reporting Methodology Evolution
  • What happened? Reporting
  • Why happened?  Analytic
  • What is happening? Monitor
  • What will happen? Prediction
  • How to resolve? Prescriptive

Visualization Methodology
4 Key factors to build visualization models
  1. Distribution
  2. Relationship
  3. Comparison
  4. Composition

Few industry wide popular analysis methods
  • Weighted average rating
  • Logistic regression
  • Time Series Analysis
  • Stationary series
  • Principal Component analysis

R Language
  • R is considered as key programming language for analytics becaz
  • Open source running on Windows,Mac,Linux,Unix,etc
  • Contains industry wide statistical package
  • Measure of dispersion
  • 3 versions: Simple R, Parallel R, Distributed R
  • Robust, vibrant community to support
  • Embedded data visualization techniques

Sector wise Demo site:
http://genisights.com/retail/home.jsp
http://genisights.com/finance/home.jsp

2 comments: