A Forecast Model Applied to Monitor Crops Dynamics Using Vegetation Indices (NDVI)

Vegetation dynamics is very sensitive to environmental changes, particularly in arid zones where climate change is more prominent. Therefore, it is very important to investigate the response of this dynamics to those changes and understand its evolution according to different climatic factors. Remote sensing techniques provide an effective system to monitor vegetation dynamics on multiple scales using vegetation indices (VI), calculated from remote sensing reflectance measurements in the visible and infrared regions of the electromagnetic spectrum. In this study, we use the normalized difference vegetation index (NDVI), provided from the MOD13Q1 V006 at 250 m spatial resolution product derived from the MODIS sensor. NDVI is frequent in studies related to vegetation mapping, crop state indicator, biomass estimator, drought monitoring and evapotranspiration. In this paper, we use a combination of forecasts to perform time series models and predict NDVI time series derived from optical remote sensing data. The proposed ensemble is constructed using forecasting models based on time series analysis, such as Double Exponential Smoothing and autoregressive integrated moving average with explanatory variables for a better prediction performance. The method is validated using different maize plots and one olive plot. The results after combining different models show the positive influence of several weather measures, namely, temperature, precipitation, humidity and radiation.

Seasonally fitted probability functions changing weights for combining vegetation indices forecasting models

<p>Normalized Difference Vegetation Index (NDVI) serves as a significant reference for crop health monitoring. NDVI time series forecasting is a critical issue because of the importance of the involving fields, e.g., food scarcity, climate changes and biodiversity. Therefore, several forecasting models have been suggested and implemented in the literature. Herein, we propose a combination of forecasts using seasonally fitted probability functions changing weights. Contrary to commonly suggested combination models, this one does not rely on overall error measures and/or features, but on time slots similarities between probability density function (PDF) of real observations and forecasts. It is validated with 18 years MOD13Q1 NDVI time series describing a cereal canopy area that belongs to the northwestern of Tunisia. Additionally, the chosen forecasting models are Box Jenkins and Neural Network model. The forecasting accuracy was assessed using the root mean square error (RMSE). According to the results, each season had a different best-fit probability distribution function. Overall, these latter are: Gamma, Beta, Weillbul, and Extreme Generalised Value (EGV). Moreover, the suggested model has shown better forecasting accuracy than individual models, hybrid models and commonly used combining tool (RMSE respectively, 0.003, 0.45, 0.35, 0.38). Interestingly, another seasonally varying weights were determined based on the normal distribution. But, our suggested model showed better forecasting accuracy than this latter (RMSE of normally distributed changing weights= 0.30).</p><div> <div> <div> </div> <div> <div> <div> </div> <div> <p> </p> <p> </p> </div> </div> </div> </div> </div>

Comparative Analysis of Modified Granger-Ramanathan and Bates-Granger Methods to Combine Forecasts of the Dynamics of Economic Indicators

Combining forecasts is one of the most effective and well-established methods for improving the accuracy of economic forecasting. This approach allows the use of all available information about the predicted phenomenon contained in individual forecasting methods. Moreover, today there are many approaches to construct weights, through which particular forecasts are combined.But with a large variety of methods for constructing weight coefficients, there are a number of problems, primarily concerning the interpretation of the weight coefficients that affect the accuracy of forecasts. The purpose of this paper is to analyze the previously proposed approaches to modify the most popular methods for constructing the weighting coefficients of Granger-Ramanathan and Bates-Granger, which allow to solve the problem of the possibility of obtaining negative weights when combining forecasts. As well as to compare the accuracy of the results when using data modifications of the methods for combining forecasts with private forecasting methods and with the original methods of combining.All the methods described in the work were used to predict some specific types of industrial products produced in Russia, presented as annual data for the period from 1952 to 2018: steel, coke, plywood and cement. Based on the developed forecasts, the accuracy of the obtained results was compared.As a result of the analysis, it was determined that the combination of forecasts remains the most effective method for improving the accuracy of forecasting, and the modifications proposed by the authors to the methods for constructing weight coefficients deserve their further use in economic practice.

A comparative study on combinations of forecasts and their individual forecasts by means of simulated series

Over the years, several studies that compare individual forecasts with the combination of forecasts were published. There is, however, no unanimity in the conclusions. Furthermore, methods of combination by regression are poorly explored. This paper presents a comparative study of three methods of combination and their individual forecasts. Based on simulated data, it is evaluated the accuracy of Artificial Neural Networks, ARIMA and exponential smoothing models; calculating the combined forecasts through simple average, minimum variance and regression methods. Four accuracy measurements, MAE, MAPE, RMSE and Theil’s U, were used for choosing the most accurate method. The main contribution is the accuracy of the combination by regression methods.

Developing and Comparing Different Strategies for Combining Probabilistic Photovoltaic Power Forecasts in an Ensemble Method

Accurate probabilistic forecasts of renewable generation are drivers for operational and management excellence in modern power systems and for the sustainable integration of green energy. The combination of forecasts provided by different individual models may allow increasing the accuracy of predictions; however, in contrast to point forecast combination, for which the simple weighted averaging is often a plausible solution, combining probabilistic forecasts is a much more challenging task. This paper aims at developing a new ensemble method for photovoltaic (PV) power forecasting, which combines the outcomes of three underlying probabilistic models (quantile k-nearest neighbors, quantile regression forests, and quantile regression) through a weighted quantile combination. Due to the challenges in combining probabilistic forecasts, the paper presents different combination strategies; the competing strategies are based on unconstrained, constrained, and regularized optimization problems for estimating the weights. The competing strategies are compared to individual forecasts and to several benchmarks, using the data published during the Global Energy Forecasting Competition 2014. Numerical experiments were run in MATLAB and R environments; the results suggest that unconstrained and Least Absolute Shrinkage and Selection Operator (LASSO)-regularized strategies exhibit the best performances for the datasets under study, outperforming the best competitors by 2.5 to 9% of the Pinball Score.

Industry-based equity premium forecasts

Purpose This paper aims to study whether the industry indexes predict the evolution of the broad stock market in the USA. Design/methodology/approach The study uses industry indexes to predict the equity premium in the USA. It considers several types of predictive models: constant coefficients and constant volatility, drifting coefficients and constant volatility, constant coefficients and stochastic volatility and drifting coefficients and stochastic volatility. The models are estimated through the particle learning algorithm, which is suitable for dealing with the problem that an investor faces in practice, given that it allows the investor to revise the parameters as new information arrives. The individual forecasts are combined based on their past performance. Findings The results reveal that models exhibit significant predictive ability. The models with constant volatility exhibit better performance, at the statistical level, but the models with stochastic volatility generate higher gains for a mean–variance investor. Practical implications This study’s findings are valuable not only for finance researchers but also for private investors and mutual fund managers, who can use these forecasts to improve the performance of their portfolios. Originality/value To the best of the knowledge of the author, this is the first paper that uses particle learning and combination of forecasts to predict the equity premium in the USA based on industry indexes. The study shows that the models generate valuable forecasts over the long time span that is considered.