Predictive ability of machine learning methods for massive crop yield prediction

Abstract

Crop yield prediction (CYP) is a major problem in agriculture. Starting each growing season, agricultural planners require estimating the yield for all the involved crops (Frausto-Solis et al., 2009). Regrettably, CYP is difficult because it depends on many interrelated factors (Liu et al., 2001;Marinković et al., 2009). Moreover, yield is also affected by farmer decisions (such as applied irrigations, pest and fertilizers applications, crop rotation, and land preparation) and incontrollable factors (such as weather, subsidies and market). As stated by Ruß (2009), yield prediction traditionally has relied on farmers' long-term experience for specific fields, crops and climate conditions, which can be inaccurate. Simple estimators, such as the average of several previous yields or the last obtained yield, are also used. Nevertheless, crop yield varies spatially and temporally with a non-linear behavior (Liu et al., 2001;Drummond et al., 2003;Schlenker & Roberts, 2006) AbstractAn important issue for agricultural planning purposes is the accurate yield estimation for the numerous crops involved in the planning. Machine learning (ML) is an essential approach for achieving practical and effective solutions for this problem. Many comparisons of ML methods for yield prediction have been made, seeking for the most accurate technique. Generally, the number of evaluated crops and techniques is too low and does not provide enough information for agricultural planning purposes. This paper compares the predictive accuracy of ML and linear regression techniques for crop yield prediction in ten crop datasets. Multiple linear regression, M5-Prime regression trees, perceptron multilayer neural networks, support vector regression and k-nearest neighbor methods were ranked. Four accuracy metrics were used to validate the models: the root mean square error (RMS), root relative square error (RRSE), normalized mean absolute error (MAE), and correlation factor (R). Real data of an irrigation zone of Mexico were used for building the models. Models were tested with samples of two consecutive years. The results show that M5-Prime and k-nearest neighbor techniques obtain the lowest average RMSE errors (5.14 and 4.91), the lowest RRSE errors (79.46% and 79.78%), the lowest average MAE errors (18.12% and 19.42%), and the highest average correlation factors (0.41 and 0.42). Since M5-Prime achieves the largest number of crop yield models with the lowest errors, it is a very suitable tool for massive crop yield prediction in agricultural planning.

Machine learning applied to the prediction of citrus production

“…MAE lowest values were obtained when production was predicted for mandarin (0.081), obtaining values around 0.10 for lemon and sweet orange production. These values were similar to those obtained in González-Sánchez et al (2014). Both RMSE and MAE were computed as the average, obtained over the 100 repetitions of the bootstrapping process.…”

“…Finally, the worst R is obtained for predicting mandarin production. All these values were good enough, compared to the values obtained in González-Sánchez et al (2014), for yield prediction. Regarding RMSE, the highest error was obtained when the production was predicted for sweet orange (0.297), whereas the lemon prediction was the most accurate (0.072).…”

“…M5-Prime is an improvement over M5 that can deal with missing values and enumerated attributes (Wang & Witten, 1997) and has been used, for instance, to predict streamflow (Onyari & Ilunga, 2013), to model sediment yield (Goyal, 2014), to estimate the maximum scour depth at breakwaters (Pourzangbar et al, 2017) and to predict the compressive strength of high performance concrete (Behnood et al, 2017). González-Sánchez et al (2014) (Ye et al, 2007).…”

Maize yield and nitrate loss prediction with machine learning algorithms

“…Traditionally, farmers rely on their experiences and past historical data such as the crop yields and weather to make important decisions to increase short-term profitability and long-term sustainability of their operation (Arbuckle and Rosman 2014). New promising technologies such as machine learning (ML) have emerged over the last years that can potentially aid farmers' decision making (Hoogenboom et al 2004, González Sánchez et al 2014, Togliatti et al 2017, Basso and Liu 2018, Ansarifar and Wang 2018, Moeinizade et al 2019. However, the lack of spatial and temporal data that cover a range of production (e.g.…”

Analysis of agriculture data using data mining techniques: application of big data

“…As an example, the k-means clustering is used for clustering soils in combination with GPSbased technologies [11]. Authors like Alberto Gonzalez-Sanchez, Juan Frausto-Solis and Waldo Ojeda-Bustamante have done extensive study on predictive ability of machine learning techniques such as multiple linear regression, regression trees, artificial neural network, support vector regression and k-nearest neighbour for crop yield production [12]. Wheat yield prediction using machine learning and advanced sensing techniques has done by Pantazi, DimitriosMoshou, Thomas Alexandridis and Abdul MounemMouazen [13].…”