Predictive Analytics

Use Cases for Big Data Analytics

Analysis for the past; Monitoring for the present; and Modeling for the future.

(3Vs for Big Data: Volume, Velocity, Variety)

Sentiment Analytics in Financial Markets: Sentiment Asset Pricing; News Sentiment (Tweets Analytics) and Market Sentiment (VIX).
Federal Market Monitoring: Data Model; Visualization; Modeling; Flash Crash Signal Detection
Performance Tuning and Sizing in System Engineering: Predict system performance (DV) with algos and benchmark data for IVs such as CPU, Memory, Design, and Parallel Processing (e.g., multithreading and clustering).

Approach
The two nonexclusive options for predictive modeling and analytics are Statistical Data Analysis and Big Data (e.g., SAPE and Queuing) Modeling. The limitation of statistical modeling is the sole reliance on existing (past) data and the lack of future representations in the statistical models (e.g., the Capital Asset Pricing Model for stock markets).

Engaging statistical modeling with data abstraction from benchmark systems, Big Data Modeling addresses the limitation of not representing future surprises in the models. For example, the Big Data Modeling approach may entail:

establish nonlinear algos to depict the relationship between benchmark (abstraction) IVs and target (future) DVs.
collect benchmark data for IVs.
collect benchmark to target ratios (e.g., enabled by performance improvements due to parallel processing).
compute target DV.

SOA Parallel Processing Design Pattern