Multisensor crop yield estimation with machine learning

2020 ◽  
Author(s):  
Laura Martínez Ferrer ◽  
Maria Piles ◽  
Gustau Camps-Valls
Keyword(s):  
Machine Learning ◽  
Time Series ◽  
Linear Regression ◽  
Crop Yield ◽  
Gaussian Processes ◽  
Vegetation Index ◽  
Maximum Temperature ◽  
Yield Estimation ◽  
Study Results ◽  
Non Linear

<p>Providing accurate and spatially resolved predictions of crop yield is of utmost importance due to the rapid increase in the demand of biofuels and food in the foreseeable future. Satellite based remote sensing over agricultural areas allows monitoring crop development through key bio-geophysical variables such as the Enhanced Vegetation Index (EVI), sensitive to canopy greenness, the Vegetation Optical Depth (VOD), sensitive to biomass water-uptake dynamics, and Soil Moisture (SM), which provides direct information of plant available water. The aim of this work is to implement an automatic system for county-based crop yield estimation using time series from multisource satellite observations, meteorological data and available in situ surveys as supporting information. The spatio-temporal resolution of satellite and meteorological observations are fully exploited and synergistically combined for crop yield prediction using machine learning models. Linear and non-linear regression methods are used: least squares, LASSO, random forests, kernel machines and Gaussian processes. Here we are not only interested in the prediction skill, but also on understanding the relative relevance of the covariates. For this, we first study the importance of each feature separately and then propose a global model for operational monitoring of crop status using the most relevant agro-ecological drivers.</p><p> </p><p>We selected the Continental U.S. and a four-year time series dataset to perform the research study. Results reveal that the three satellite variables are complementary and that their combination with maximum temperature and precipitation from meteorological stations provides the best estimations. Interestingly, adding information about crop planted area also improved the predictions. A non-linear regression model based on Gaussian processes led to best results for all considered crops (soybean, corn and wheat), with high accuracy (low bias and correlation coefficients ranging from 0.75 to 0.92). The feature ranking allowed understanding the main drivers for crop monitoring and the underlying factors behind a prediction loss or gain.</p>

Forecasting stock market price by using fuzzified Choquet integral based fuzzy measures with genetic algorithm for parameter optimization

2020 ◽  
Vol 54 (2) ◽  
pp. 597-614
Author(s):  
Shanoli Samui Pal ◽  
Samarjit Kar
Keyword(s):  
Genetic Algorithm ◽  
Time Series ◽  
Linear Regression ◽  
Regression Model ◽  
Choquet Integral ◽  
Time Series Forecasting ◽  
Fuzzy Measure ◽  
Model Parameters ◽  
Forecasting Models ◽  
Non Linear

In this paper, fuzzified Choquet integral and fuzzy-valued integrand with respect to separate measures like fuzzy measure, signed fuzzy measure and intuitionistic fuzzy measure are used to develop regression model for forecasting. Fuzzified Choquet integral is used to build a regression model for forecasting time series with multiple attributes as predictor attributes. Linear regression based forecasting models are suffering from low accuracy and unable to approximate the non-linearity in time series. Whereas Choquet integral can be used as a general non-linear regression model with respect to non classical measures. In the Choquet integral based regression model parameters are optimized by using a real coded genetic algorithm (GA). In these forecasting models, fuzzified integrands denote the participation of an individual attribute or a group of attributes to predict the current situation. Here, more generalized Choquet integral, i.e., fuzzified Choquet integral is used in case of non-linear time series forecasting models. Three different real stock exchange data are used to predict the time series forecasting model. It is observed that the accuracy of prediction models highly depends on the non-linearity of the time series.

scholarly journals Attacking Data Transforming Learners at Training Time

2019 ◽  
Vol 33 ◽  
pp. 3167-3174
Author(s):  
Scott Alfeld ◽  
Ara Vartanian ◽  
Lucas Newman-Johnson ◽  
Benjamin I.P. Rubinstein
Keyword(s):  
Machine Learning ◽  
Time Series ◽  
Linear Regression ◽  
Ordinary Least Squares ◽  
General Purpose ◽  
Training Time ◽  
Target Values ◽  
Gradient Based ◽  
Deployment Time ◽  
Compositional Learning

While machine learning systems are known to be vulnerable to data-manipulation attacks at both training and deployment time, little is known about how to adapt attacks when the defender transforms data prior to model estimation. We consider the setting where the defender Bob first transforms the data then learns a model from the result; Alice, the attacker, perturbs Bob’s input data prior to him transforming it. We develop a general-purpose “plug and play” framework for gradient-based attacks based on matrix differentials, focusing on ordinary least-squares linear regression. This allows learning algorithms and data transformations to be paired and composed arbitrarily: attacks can be adapted through the use of the chain rule—analogous to backpropagation on neural network parameters—to compositional learning maps. Bestresponse attacks can be computed through matrix multiplications from a library of attack matrices for transformations and learners. Our treatment of linear regression extends state-ofthe-art attacks at training time, by permitting the attacker to affect both features and targets optimally and simultaneously. We explore several transformations broadly used across machine learning with a driving motivation for our work being autogressive modeling. There, Bob transforms a univariate time series into a matrix of observations and vector of target values which can then be fed into standard learners. Under this learning reduction, a perturbation from Alice to a single value of the time series affects features of several data points along with target values.

scholarly journals Prediction and Forecasting of Air Quality Index in Chennai using Regression and ARIMA time series models

2021 ◽  
Vol 9 ◽  
Author(s):  
Geetha Mani ◽  
◽  
Joshi Kumar Viswanadhapalli ◽  
Albert Alexander Stonie ◽  
◽  
...  
Keyword(s):  
Machine Learning ◽  
Time Series ◽  
Air Quality ◽  
Linear Regression ◽  
Quality Index ◽  
Air Quality Index ◽  
Model Parameters ◽  
Sensor Output ◽  
Model Accuracy ◽  
Life On Earth

Air is one of the most fundamental constituents for the sustenance of life on earth. The meteorological, traffic factors, consumption of non-renewable energy sources, and industrial parameters are steadily increasing air pollution. These factors affect the welfare and prosperity of life on earth; therefore, the nature of air quality in our environment needs to be monitored continuously. The Air Quality Index (AQI), which indicates air quality, is influenced by several individual factors such as the accumulation of NO2, CO, O3, PM2.5, SO2, and PM10. This research paper aims to predict and forecast the AQI with Machine Learning (ML) techniques, namely linear regression and time series analysis. Primarily,Multi Linear Regression (MLR) model, supervised machine learning, is developed to predict AQI. NO2, Ozone(O3), PM 2.5, and SO2 sensor output collected from Central Pollution Control Board (CPCB) – Chennai region, India feed as input features and optimized AQI calculated from sensor's output set as a target to train the regression model. The obtained model parameters are validated with new and unseen sensor output. The Key Performance Indices(KPI) like co-efficient of determination, root mean square error and mean absolute error were calculated to validate the model accuracy. The K-cross-fold validation for testing data of MLR was obtained as around 92%. Secondly, the Auto-Regressive Integrated Moving Average (ARIMA) time series model is applied to forecast the AQI. The obtained model parameters were validated with unseen data with a timestamp. The forecasted AQI value of the next 15 days lies in a 95 % confidence interval zone. The model accuracy of test data was obtained as more than 80%.

scholarly journals Estimating the Effect of Climatic Variables and Crop Area on Maize Yield in Ghana

2012 ◽  
Vol 3 (9) ◽  
pp. 313-321
Author(s):  
Henry De-Graft Acquah
Keyword(s):  
Climate Change ◽  
Linear Regression ◽  
Regression Model ◽  
Crop Yield ◽  
Crop Production ◽  
Growing Season ◽  
Maize Yield ◽  
Climatic Variables ◽  
Non Linear ◽  
Crop Area

Climate change tends to have negative effects on crop yield through its influence on crop production. Understanding the relationship between climatic variables, crop area and crop yield will facilitate development of appropriate policies to cope with climate change. This study therefore examines the effects of climatic variables and crop area on maize yield in Ghana based on regression model using historical data (1970-2010). Linear and Non-linear regression model specifications of the production function were employed in the study. The study found that growing season temperature trend is significantly increasing by 0.03oC yearly whereas growing season rainfall trend is insignificantly increasing by 0.25mm on yearly basis. It was also observed that rainfall is becoming increasingly unpredictable with poor distributions throughout the season. Results from the linear and non-linear regression models suggest that rainfall increase and crop area expansion have a positive and significant influence on mean maize yield. However, temperature increase will adversely affect mean maize yield. In conclusion, the study found that there exists not only a linear but also a non-linear relationship between climatic variables and maize yield.

scholarly journals ML Methods for Crop Yield Prediction and Estimation: An Exploration

2020 ◽  
Vol 9 (3) ◽  
pp. 3506-3508
Keyword(s):  
Machine Learning ◽  
Detailed Analysis ◽  
Crop Yield ◽  
Learning Process ◽  
Machine Learning Techniques ◽  
Yield Prediction ◽  
Yield Estimation ◽  
Data Set ◽  
Yield Optimization ◽  
Learning Techniques

Machine learning Has performed a essential position within the estimation of crop yield for both farmers and consumers of the products. Machine learning techniques learn from data set related to the environment on which the estimations and estimation are to be made and the outcome of the learning process are used by farmers for corrective measures for yield optimization. This paper we explore various ML techniques utilized in crop yield estimation and provide the detailed analysis of accuracy of the techniques.

scholarly journals Time Series Analysis of MODIS-Derived NDVI for the Hluhluwe-iMfolozi Park, South Africa: Impact of Recent Intense Drought

2018 ◽  
Author(s):  
Nkanyiso Mbatha ◽  
Sifiso Xulu
Keyword(s):  
Time Series ◽  
Vegetation Index ◽  
Southern Oscillation ◽  
Kendall Test ◽  
Google Earth ◽  
Maximum Temperature ◽  
Sudden Increase ◽  
Area Index ◽  
Enhanced Vegetation Index ◽  
Temperature And Precipitation

The variability of meteorological parameters such as temperature and precipitation, and climatic conditions such as intense droughts, are known to impact vegetation health over southern Africa. Thus, understanding large-scale ocean–atmospheric phenomena like the El Niño/Southern Oscillation (ENSO) and Indian Ocean Dipole/Dipole Mode Index (DMI) is important as these factors drive the variability of temperature and precipitation. In this study, 16 years (2002–2017) of Moderate Resolution Imaging Spectroradiometer (MODIS) Terra/Aqua 16-day normalized difference vegetation index (NDVI), extracted and processed using JavaScript code editor in the Google Earth Engine (GEE) platform in order to analyze the response pattern of the oldest proclaimed nature reserve in Africa, the Hluhluwe-iMfolozi Park (HiP), during the study period. The MODIS-enhanced vegetation index and burned area index were also analyzed for this period. The area-averaged Modern Retrospective Analysis for Research Application (MERRA) model maximum temperature and precipitation were also extracted using the JavaScript code editor in the GEE platform. This procedure demonstrated a strong reversal of both the NDVI and Enhanced Vegetation Index (EVI), leading to signs of a sudden increase of burned areas (strong BAI) during the strongest El Niño period. Both the Theilsen method and the Mann–Kendall test showed no significant greening or browning trends over the whole time series, although the annual Mann–Kendall test, in 2003 and 2014–2015, indicated significant browning trends due to the most recent strongest El Niño. Moreover, a multi-linear regression model seems to indicate a significant influence of both ENSO activity and precipitation. Our results indicate that the recent 2014–2016 drought altered the vegetation condition in the HiP. We conclude that it is vital to exploit freely available GEE resources to develop drought monitoring vegetation systems, and to integrate climate information for analyzing its influence on protected areas, especially in data-poor counties.

Sign in / Sign up

Export Citation Format

Share Document