On optimal levels of influential variables for the growth of with directional data
by
S. Bhattacharya
Indian Statistical Institute, Calcutta, India
Coauthors: A. SenGupta and T. Basu (Indian Statistical Institute, Calcutta, India)

A real-life experiment was conducted over a period of about two years under a circular design to incorporate the spatio-temporal variability of several influential variables on the growth of fish. The standard Q-Q plots suggest the non-normality of growth data of one Indian major carp Cirrhinus mrigala. Similar plots for two factors DO (Dissolved Oxygen) and TMP (Temperature), which are responsible for the growth of fish, reveal the same phenomenon. So standard ANOVA test is not applicable here. Certain transformations have been suggested to yield proximity to normal distribution for this purpose. We next determine the regression equations with one and both the explanatory variables DO and TMP to predict the growth of the fish. The directional variations in the above explanatory variables were found to be significant and a model is suggested to incorporate this fact. Exploiting this model, the optimal levels of DO and TMP are determined which yield the maximum growth of the fish.

Date received: November 13, 2000