Insight into the history of the development of statistical data analysis methods in the research of fruit growing in Latvia

Throughout the 20^th century, statistical data analysis in fruit-growing research, particularly in field trials, has been closely related to the practice in agricultural science in Latvia.
Initially, with the simple and straightforward implementation of standard methods, a trial is established by placing all replications of the treatment in a single row, and methods of direct comparison are used until more complex factorial ones are employed.
The data analysis of that time was related to the technical possibilities of computing.
It limited multifactorial investigations and gradations of factors in the trials.
Dr.
Agr.
Ivars Dimza, together with colleagues, solved limitations with methods of the ternary grid, fractional replication methods, etc.
There was a place for discussion of the level of critical significance, 0.05. Trials in the orchards related to uneven soil conditions, the necessity to prove interactions of factors, as well as the longevity of fruit crops staying in one place, with the cumulative effect of conditions.
Due to the development of computing technologies, ANOVA and regression analysis are now possible.
The data from the split-plot and randomized block design trials were analyzed and programmed for use.
With multiple regression analysis, it was possible to combine the data of thematically similar trials.
It allowed for the analysis of the influence and interactions of qualitative and quantitative factors, estimating the optimum of models with quadratic and higher-degree nonlinear equations.
Coding and centering of factors (arithmetic means of them) and estimated deviations of data were key to applying modeling equations, such as predicting optimum concentrations of investigated nutrients, etc.
Even breeding material and cultivars with qualitative parameter characters can be used for data analysis, where they are ranked by certain parameter indications and then coded and centered.
In fruit science, it is the basis of future data analysis, taking into account the perennials.

EHC2024: International Symposium on the History of Horticulture in Europe

E. Rubauskis, J. Lepsis

multiple regression analysis, data coding, centring, perennial crops

https://doi.org/10.17660/ActaHortic.2025.1438.56