Nauka Przyroda Technologie

Science Nature Technologies

Scientific journal presents original papers and articles relevant to all aspects of agricultural sciences.

Język polski English
Quick search:

2018 volume 12 issue 3, 285–296
Gniewko Niedbała, Magdalena Piekutowska, Tomasz Piskier, Mariusz Adamski, Robert Korzeniewicz
Application of the multiple regression method to predict the quality characteristics of potato tubers of the Innovator variety

Introduction. The potato is a common species in cultivation, mainly on the European and Asian continents. It is estimated that total production in Asia and Europe accounts for 85% of world production. Potato production for French fries differs from the cultivation of ware potatoes or starch potatoes. One of the most important factors determining the success of cultivation is varie-ty. Potato growers who grow potatoes for French fries usually choose foreign varieties. An important aspect of specialist agricultural production is the possibility of forecasting the expected yields in terms of quantity, but also quality. On the basis of data on the rate of dry matter growth in tubers during the growing season, farmers can plan in advance the date of defoliation or excavation. The aim of this study was to obtain a regression model predicting the parameter of tuber weight underwater (UWW – underwater weight) for Innovator potato variety. Knowing its value it is easy to determine the dry matter and starch content of tubers.
Materials and methods. For the construction of the regression model forecasting UWW Innova-tor cultivar, data from the production fields of the individual farm from 2011–2017 were used. All fields were located in the northern Poland, in the Słupski and Sławieńki districts (Pomeranian and West Pomeranian Voivodeships). The Dutch potato variety Innovator (HZPC cultivator) is one of the most popular varieties intended for French fries. In each year of cultivation of the above mentioned cultivar, the number of fields and the area planted with potatoes was different in the farm in question. The year with the smallest scale of production was the year 2012 – 3 fields, 204 ha, while the highest number of potatoes was cultivated in 2017 – 5 fields, 344 ha. In order to build a regression model, which is the subject of this study, data from the last, September samples in the years 2011–2017 were used. In total, results from 82 samples taken from the field were worked on. The set for the construction of the regression model, called set I, consisted of 75 samples. Set II, which consisted of 7 randomly selected samples, performed a validation function and did not participate in the construction of the model. The structure of the model is based on 8 independent features – meteorological data and mineral fertilization levels.
Results. The produced regression model is characterized by a determination coefficient R2 = 0.6623. At the level of statistical significance α = 0.05 four independent features influencing UWW to the greatest extent were determined. These are: average air temperature from 1st April to the day in September when the sample was taken (T4-9), sum of N fertilization in the current year (N), sum of SO3 fertilization in the current year (SO3) and sum of MgO fertilization in the current year (MgO). Four ex-post error measures were used for model validation, i.e. global relative approximation error of the model (RAE = 0.0382), root mean square error (RMS = 22.5781), mean absolute error (MAE = 14.5978), mean absolute percentage error (MAPE = 3.823).
Conclusions. The produced regression model is characterized by a low value of the mean absolute percentage error (MAPE) of 3.823%. This means that it can be used in agricultural practice. The factor with the highest weight affecting the dry weight of potato tubers on the basis of the weight of tubers under water (Y_UWW) is the average daily temperature (T4-9) in the months from 1st April to x September.

Key words: prediction, regression model, potato, Innovator, dry matter
PDFFull text available in Polish in Adobe Acrobat format:

For citation:

MLA Niedbała, Gniewko, et al. " Zastosowanie metody regresji wielorakiej do predykcji cech jakościowych bulw ziemniaka odmiany ‘Innovator’." Nauka Przyr. Technol. 12.3 (2018): 285–296.
APA Gniewko Niedbała1, Magdalena Piekutowska2, Tomasz Piskier2, Mariusz Adamski1, Robert Korzeniewicz3 (2018). Zastosowanie metody regresji wielorakiej do predykcji cech jakościowych bulw ziemniaka odmiany ‘Innovator’. Nauka Przyr. Technol. 12 (3), 285–296
ISO 690 NIEDBAłA, Gniewko, et al. Zastosowanie metody regresji wielorakiej do predykcji cech jakościowych bulw ziemniaka odmiany ‘Innovator’. Nauka Przyr. Technol., 2018, 12.3: 285–296.
Corresponding address:
Gniewko Niedbała
Instytut Inżynierii Biosystemów
Wydział Rolnictwa i Bioinżynierii
Uniwersy-tet Przyrodniczy w Poznaniu
ul. Wojska Polskiego 50
60-627 Poznań,Poland
Accepted for print: 30.11.-0001