Consultancy Statistics

Consultancy Modeling en Optimization

Trendanalist is an unique software program to perform trend analysis on data series. It is a statistical technique to detect objectivily whether a trend exist in a data series and to estimate the trend value. In Trendanalist are some specific statistical techniques implemented, various linear modelling methods (ao Box-Jenkins methodology) and three variants of the Mann-Kendall test (with or without autocorrelation and seasonal effects). The program can handle specific characteristics of data series of environmental variables (especially water quality data) like: skewed or non-normal distribution, seasonal effects, autocorrelation, missing values, censored values and extreme values. Trendanalist has an easy to use interface.

Trendanalist contains an expert system to perform the trend analysis automatically. The expert systen follows a 'streaming scheme', so that the trend analysis and the generation of a report of a hundred data series - of for instance a monitoring network of water quality - last a few minutes. The expert system always chooses the optimal trend test with the maximum power and minimal risc of erroneous trend detection. The user can always analyse individual data series manually, because Trendanalist contains an extensive statistical toolbox. In the user guide you find information about the possibilities and backgrounds of Trendanalist.

Trendanalist now presents along with the trend a modelling of the time serie with Lowess. Lowess (locally weighted scatterplot smooth) is a modelling with polynomials of parts of the time serie for a smooth graphical description of the data. The comparison of the trend and the Lowess line gives complementary information for the judgement of a change in a time serie.

The download of Trendanalist is accessibled with a code. Fill in the code and click on submit (no 'Enter') to activate the code.