Consultancy Modelling and Optimization

TRG ('Tijdreeksanalist') is an easy to use transparant software program to analyze and model time series to make simulations and predictions. In TRG the Box-Jenkins modelling method and extensions is implemented. TRG has been developed in the software environment of Matlab. We deliver the Matlab or the standalone version of TRG.

In the Netherlands experts advice TRG. TRG gives you the right possibilities to detect the appropiate time series model out of a wide range of possible time series models to give a maximum likehood description between the process variables. TRG has been developed to perform time series analysis for everyone with any knowledge of statistics and modelling, because TRG is in that way transparant (not a black box) and prevents users to chooce the wrong way and get bad modelling results. TRG is simple and logic with a lot of functionalities, like:

• The general Box-Jenkins and ARIMA time series modelling with some extensions
• Data transformation
• Data analysis
• A smooth method to detect outliers
• Correlation analysis
• Advanced and interactive parameter estimation methods
• Parameter analysis ans reduction methods
• Residual analysis
• Simulation and prediction possibilities

New in version 4.0 is the possibility to model an unrestrained number of time series with a wide range of time series models and to optimize them according to a specified criterion. TRG exports a report and if necessary the models can afterwards manually be analyzed and optimized with the extensive toolbox in TRG. These new functionality gives a huge eit geeft een grote tijdsbesparing in de analyse van grote meetnetten.

New in version 4.1 is the possibility to give input variables (influences) comparible transfer functions. In this way a factor ('gewasfactor') can be estimated between inputs. In the Netherlands these factor is estimated to understand the difference between the influence of precipitation and rainfall. This option in modelling has some practical advantages.

TRG has furthermore extensive file import and file export possibilities and all analyses are graphically supported to make the time series analysis as transparant as possible.

TRG is since the first version in 1995 used in many different research areas. At the moment we (AMO-Icastat) use TRG for modelling and analyzing groundwater levels, depth measurements on the Dutch coast and traffic measurements. Recent researches with TRG are:

• 'Schets van de nieuwe grondwatersituatie in het westelijk duingebied van Zuid-Kennemerland'; water management company PWN (Noord-Holland, The Netherlands)

• 'Evaluatie hydrologische effecten stopzetten grondwaterwinning PWN in Zuid-Kennemerland'; water management company PWN and province Noord-Holland

• 'Ontwikkeling van een verklaringsmodel voor files op het traject Almere-Diemen'; AVV (Adviesdienst Verkeer en Vervoer)

• 'Tijdreeksanalyse toegepast op verkeersgegevens van de A27'; AVV

On base of the time series analysis we develop also neural networks. The time series modelleing give an accurate analysis and with a non-linear neural network there is always some gain in the description of a model of the data and possibly in the predictions of a system.

Interested?
A temporary and free license is possible to try out TRG.

The graphics below are examples of how nice the description of a time series model is of the groundwater levels in the dunes in Katwijk aan Zee (The Netherlands, Zuid-Holland). The second graph gives an estimation of the contribution of the individual influences, rainfall and waterdischarge.