Parametric modeling and simulation of complex systems using large datasets and heterogeneous data structures
A system for predicting future outcomes of dynamic and complex systems using simulation results driven by a parametric and blended analytic and modeling approach. A model engine and simulation engine in combination with a visualization engine using such an approach has been developed to produce geospatial and temporal context aware system models for use in generating predictive results which may be used to recommend future outcomes from continuously competing models derived from ingesting large amounts of varied but related data.
1. A system for predicting complex system outcomes using a multi-model, blended analysis methodology, comprising:
a model engine comprising at least a plurality of programming instructions stored in the memory of, and operating on at least one processor of, a computing device, wherein the plurality of programming instructions, when operating on the processor, cause the computing device to:
receive a request for prediction of a future outcome for a complex system based on a model input;
retrieve historical data for the complex system, wherein the historical data comprises a plurality of past events occurring within the complex system and a past outcome of the complex system as a result of the plurality of past events;
process the historical data through a machine learning algorithm that has been trained to identify correlations between the past events and the past outcome; and
generate a plurality of competing system models from the identified correlations, wherein:
each competing system model is a representation of the complex system;
each past event is represented in each competing system model by one or more individual model actors; and
the parameters of at least one individual model actor in each competing system model differ from the parameters of all other individual model actors in all other competing system models; and
a simulation engine comprising at least a plurality of programming instructions stored in the memory of, and operating on at least one processor of, a computing device, wherein the plurality of programming instructions, when operating on the processor, cause the computing device to:
retrieve the plurality of competing system models from the model engine;
simultaneously run each of the plurality of competing system models to determine a set of causal relationships between the past events and the past outcome; and
predict a future outcome of the complex system based on the model input using the causal relationships; and
a visualization engine comprising at least a plurality of programming instructions stored in the memory of, and operating on at least one processor of, a computing device, wherein the plurality of programming instructions, when operating on the processor, cause the computing device to:
display a visualization of the predicted future outcome.
2. The system of claim 1 , wherein the future outcome is based on a plurality of actual historical events from the past or from a plurality of hypothesized historical events.
3. The system of claim 1 , wherein the historical data comprises sensor data.
4. The system of claim 1 , further comprising a web-scraper.
5. The system of claim 4 , wherein the web-scraper retrieves Internet data related to the plurality of past events.
6. The system of claim 5 , wherein the related Internet data is past comprises the past outcome.
7. The system of claim 1 , wherein the machine learning algorithm is a neural networks.
8. The system of claim 1 , wherein the visualization engine renders the future outcome as one or more graphs.
9. A method for predicting complex system outcomes using a multi-model, blended analysis methodology, comprising the steps of:
using a model engine operating on a computing device comprising a memory and a processor:
receiving a request for prediction of a future outcome for a complex system based on a model input;
retrieving historical data for the complex system, wherein the historical data comprises a plurality of past events occurring within the complex system and a past outcome of the complex system as a result of the plurality of past events;
process the historical data through a machine learning algorithm that has been trained to identify correlations between the past events and the past outcome; and
generating a plurality of competing system models from the identified correlations, wherein:
each competing system model is a representation of the complex system;
each past event is represented in each competing system model by one or more individual model actors; and
the parameters of at least one individual model actor in each competing system model differ from the parameters of all other individual model actors in all other competing system models;
using a simulation engine operating on the computing device:
retrieving the plurality of competing system models from the model engine;
performing run each of the plurality of competing system models to determine a set of causal relationships between the past events and the past outcome;
predict a future outcome of the complex system based on the model input using the causal relationships; and
using a visualization engine operating on the computing device:
displaying a visualization of the predicted future outcome.
10. The method of claim 9 , wherein the future outcome is based on a plurality of actual historical events from the past or from a plurality of hypothesized events.
11. The method of claim 9 , wherein the historical data comprises sensor data.
12. The method of claim 9 , further comprising a web-scraper.
13. The method of claim 12 , wherein the web-scraper retrieves Internet data related to the plurality of events.
14. The method of claim 13 , wherein the related Internet data is past observations and past outcomes.
15. The method of claim 9 , wherein the machine learning algorithm is a neural networks.
16. The method of claim 9 , wherein the visualization engine renders the future outcome as one or more graphs.