An overview of the examples that come with the software package. Each Excel file contains data and models for quick evaluation and testing. Download our Free version and check them out.

Climate/ Environment/ Health

COD concentration Problem: Water pollution Task: Modeling, prediction Models contained: several analytical models (auto-regressive, static and dynamic input/ output models); Similar Patterns Note: This document is used by the tutorial. What Drives Global Warming? Problem: How fast is the Earth getting warmer and what are the driving forces? Task: Modeling and prediction of global warming by ozone concentration, aerosols, clouds, sun activity, and atmospheric CO2. Models contained: 8 dynamic input-output models one combined model of global warming. This example is related to a post at the Global Warming Prediction Project site. Temperature Anomalies Prediction of 9 Latitudinal Bands Problem: Is the Earth getting warmer and where and how fast? Task: Modeling and prediction of land air and sea surface temperature anomalies 36 months ahead. Models contained: 21 dynamic input-output models each built from over 4000 potential inputs and 1400 samples of historical observed temperature data. All models together establish an interdependent system of equations in the Excel file. This example is taken from the Global Warming Prediction Project site. Sunspot Number Prediction Problem: How active the sun will be in the current and the next cycle? Task: Monthly prediction 20 years ahead! Models contained: Predictions obtained by Similar Patterns modeling technology. Reproductive Toxicity of Chemical Compounds Problem: How reprotoxic is a chemical compound given its molecular structure? This is an essential question in official classification and authorization procedures of chemicals (for example REACH) to minimize and substitute the use of animal testing. Task: Modeling and prediction of developmental toxicity of chemical compounds from several hundred molecular descriptors. So-called QSAR modeling (Quantitative Structure-Activity Relationship modeling). Models contained: Static, nonlinear regression models.

Energy/ Economy/ Environment

Prediction Scenarios for Crude Oil Price Problem: How is the crude oil price developing depending on exploitation of new oil reserves and world oil consumption? Task: Simulation and prediction of yearly oil prices and oil consumption till 2025. A difficult modeling problem from a very small data set (4 inputs, 26 samples). A "high-dimensional" modeling problem, though. Models contained: Dynamic, non-linear regression models. The lesson to learn. Primary Energy Consumption by Continents Problem: Where is world energy consumption heading in the coming 10 years? Task: Forecast till 2020 by continent. Models contained: Similar Patterns modeling technology. San Francisco Solar Energy Incentive Program Problem: To encourage more installations of solar power, the City has been offering incentives to residents and businesses to install solar power on their properties. It empowers residents, business owners and non-profits to reduce their own carbon footprints and power. Task: Installed solar power prediction 6 months ahead. Again, a very small data set (5 inputs, 26 samples). Models contained: Static, non-linear regression model and Similar Patterns. Global CO2 Emission by Energy Consumption Problem: How worldwide CO2 emissions are driven by different types of energy consumption (nuclear, oil, gas, coal, hydro)? Task: Modeling and forecasting average yearly emissions till 2025. Models contained: Static, non-linear regression model and Similar Patterns. BP Statistical Review of World Energy June 2010 Task: Reference data source.

Engineering/ Other

Computer System Activity of a multi-processor, multi-user system Problem: Modeling, i.e., finding the relationship that describes system activity from a number of usage variables. Task: Predictive modeling and identification. Models contained: Static, non-linear regression model. Laser System State Problem: Modeling and prediction of a laser system over time. Task: One-step prediction for the next 300 steps in time. Models contained: Non-linear auto-regressive model. Housing Value Estimation Problem: Finding a model that objectively estimates housing values of a region. Task: Modeling, Interpretation Models contained: Non-linear regression model Boiling Points of Gases Problem: Boiling points estimation as a basic chemical property. Task: Modeling, prediction Models contained: Static, non-linear regression model. Very small data set (3 inputs, 8 samples).

Q & A

Can KnowledgeMiner (yX) self-organize the true model? Problem: Artificial model identification problem. A noise free and a noisy data set is given. The equation that was used to generate this data is known to the user in this case. Is KnowledgeMiner able to identify this equation by looking at the data only? Also for noisy data? A basic problem of data-driven modeling. Task: Model identification. Models contained: Non-linear auto-regressive regression models. Model Accuracy and Noisy Data Problem: A small related study on the general problem of noise filtering and noise immunity of a modeling method. Task: Predictive Modeling and Generalization. Models contained: Linear and non-linear regression models