Thumbnail Image

Theses

Browse

19902
Reset filters
Subject: Agricultural Statistics and Informatics × Author: KAU × End date: 1990 × Start date: 1990 × Thesis Type: M.Sc ×
Reset filters

Search Results

Now showing 1 - 2 of 2
  • Thumbnail Image
    ThesisItemOpen Access
    Influence of weather parameters on the yield of black pepper
    (Department of Agricultural Statistics, College of Horticulture, Vellanikkara, 1990) Dessy Mabel; KAU; Gopinathan Unnithan, V K
    Influence of weather parameters on the yield of black pepper was studied utilising the data on yield of 29 varieties of pepper (Pipger nigrum) and maximum temperature, minimum temperature, relative humidity, rainfall and number of rainy days recorded from 1963-64 to 1979-80 at The Pepper Research Station, Kerala Agricultural University, Panniyoor, in the Cannanore District of Kerala. Averages/totals of weather elements for standard weeks as well as for fortnights during the critical period of crop growth viz, from April 9 to August 12 were determined. Correlation coefficients of all the weather elements of standard weeks and fortnights with annual yield when represented in correlograms revealed, wide variations in response of different varieties to changes in climatic factors among the 29 varieties. In other wards there was definite evidence of genotype - environment interaction. Forcasting models, based on weekly as well as fortnightly weather elements were estimated for each of the 29 varieties of black pepper by two stage linear regression technique. First stage models were estimated by multiple linear regression and the second stage models were estimated with the estimates of yield from first stage models as explanatory variables by step-wise regression technique. The forecasting models utilising weekly climatic data had higher predictability compared to that utilising fortnightly data. All the final forecasting models with weekly data had predictability of 98% and above. More over the first stage models, from weekly data, could also be used to forecast yield of all varieties of black pepper with remarkable accuracy.
  • Thumbnail Image
    ThesisItemOpen Access
    Statistical investigations on the analysis of data of long term manurial trials on paddy
    (Department of Agricultural Statistics, College of Horticulture, Vellanikkara, 1990) Rani John, V; KAU; Prabhakaran, P V
    The suitability of different statistical techniques for the analysis of data of long term fertiliser trials was examined with the help of secondary data gathered from the permanent manurial experiment in paddy at Regional Agricultural Research station. Pattambi and certain new methods with distinct advantages over the existing methods were suggested for the same. The relative efficiencies of various mathematical functions in representing the yield-fertiliser relationship and in estimating the optimum level of the applied nutrient were also evaluated on the basis of secondary data gathered from the various fertiliser experiments on paddy conducted at the various rice research stations under the KAU during the last ton years. Two new mathematical.functions were also developed to represent toe response pattern for certain types of trivial data. The methods evaluated for the analysis of data of long term trials include method of groups of experiments split plot analysis* principal component analysis non parametric method proposed by Rai and Rao stability analysis proposed by Eberhart and Russell.Iran parametric stability analysis proposed by Kassar and Kuhn and analysis based on principle of game theory. A new non parametric method as an extension of Friedman's two way analysis of variance by ranks was also developed for the analysis of such data. This method was found to be almost as powerful as the method proposed by Rai and Rao and hence can be regarded as on improvement over the existing methods as it is free from any stringent assumptions on the nature of the underlying universe. Principal component analysis was also found to be empirically atleast as efficient as tho method of groups of experiments/split plot analysis and can be adjudged to be a bettor alternative to the solution of the same problem on the grounds of theoretical and statistical validity.