Hi all. I have run a non linear regression using Gompertz model on data about Conductivity and Groundwater Depth. I am uncertain on how to report this in an article and how to obtain the equation for the resulting sigmoidal curve. Here`s an image with the graph and the outcome of the statistical analysis. Questions: Do you think this model is adequate for our data? Can you help me understand the outcome of the non-linear regression? And third can we find an equation for our curve?
Dear sir you can able usediffereet types of regrission and correlation for example linear ,polynomial , Quadratic , expontial and log methods from R2 value of them you can select the best one the higher coefficent of determination (Square R = R 2 )value referes to the best model.
The equation for the curve is easy. The output tells you the estimated values of the parameters under Estimación. The equation is alpha*exp(-beta*exp(-gamma*Depth)), so it is Conductivity = 61.41 * exp(-2.87 * exp(-1.09 * Depth)).
For reporting, it is helpful if there is a p-value for the model overall and some kind of pseudo r-square value. I don't see these on your output.
In terms of adequacy, you might look at the distribution of the residuals from the analysis. I don't know if your software will provide that. Similar to an OLS model †, you want to be sure there aren't systematic errors in the analysis. For example, in your model, it appears that most of the points are below the line at small depths, and above the line at large depths. This suggests to me that the model isn't perfectly suited to capture the curvature of the data.
_________
† Ordinary least squares, such as common linear regression.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs