Time Series Forecasting of Price of Agricultural Products Using Hybrid Methods

Purohit, Sourav Kumar; Panigrahi, Sibarama; Sethy, Prabira Kumar; Behera, Santi Kumari

doi:10.1080/08839514.2021.1981659

Cited by 46 publications

(17 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…From Cenas (10) , the combination of the ARIMA and Kalman filter has better predictive performance for Philippine crop prices compared to the standalone ARIMA. According to Purohit et al (3) and Celma & Oliveira (15,16) , hybrid methods do provide better results in crop price forecasting than individual models. The result of this study also follows the same conclusion.…”

Section: Resultsmentioning

confidence: 99%

“…Researchers also tried to integrate machine learning methods to improve the prediction performance of traditional time series techniques. Integrating artificial neural networks (ANN), long short-term memory (LSTM), and support vector machine (SVM) to ETS and ARIMA have yielded superior results than the traditional models alone, even for non-Gaussian data and data having nonlinearly correlated errors (2,3) . However, the problem with these traditional and hybrid univariate time series methods is that they only consider previous values as bases for the modeling, that is, explanatory/predictor variables were not considered.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Estimation of Farmgate Rice Price Using Temporal Causal Modeling and Kalman Filtering

Patacsil¹,

Cenas²,

Jacobe³

et al. 2022

IJST

View full text Add to dashboard Cite

Objectives:To develop a forecasting model for the farmgate prices of rice crop in the Philippines and to improve the derived model forecasts by applying Kalman filters. Methods: The researcher's utilized monthly rice farmgate price and inflation rate from 1990 to 2015 as training information in building the temporal-causal model. On the other hand, the dataset for rice farmgate price from 2016 to 2020 acted as the testing set, allowing the researchers to determine model accuracy using mean square error, mean absolute error, and mean absolute percentage error. Findings: Results indicate that applying Kalman filters to the derived temporal-causal model indeed improves prediction performance, as evidenced by the lower MSE values. In particular, applying Kalman filter to the derived Temporal-Causal model 15% (without inflation as control input) and 3% (with inflation as control input) decrease in the MSE. In terms of the Temporal-Causal-Kalman filter with no control input, a decrease of 1.8% is observed for the MAE as well as a decrease of 3% in the MAPE, indicating a substantial improvement in the accuracy of the base model. Interestingly though, adding a control-input variable in the Kalman filter generated gave an increase of 4.4% and 3.8% in the MAE and MAPE respectively. This might be due to the not-so-strong correlation between the farmgate price and control input (inflation). Seeking conditions when will external inputs be helpful in enhancing Kalman filters as well as combining the models with other data analytics techniques may be valuable in future research. As for the comparison with nonlinear setups, results for unscaled, partially scaled, and fully scaled artificial neural networks show that Kalman filtering can attain almost on-par prediction performance with such methods. Novelty: This research presented a new scheme of predicting farmgate price. Compared to typical time series models, the derived Temporal-Causal model included inflation as a factor. Moreover, the combination of Temporal-Causal and Kalman filter is a new method for improving forecasts.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Estimation of Farmgate Rice Price Using Temporal Causal Modeling and Kalman Filtering

Patacsil¹,

Cenas²,

Jacobe³

et al. 2022

IJST

View full text Add to dashboard Cite

show abstract

“…Unfortunately, most agricultural price forecasting publications focus on meat, vegetables, grains, etc. [ 27 – 29 ]. Therefore, we looked for model types and functional characteristics that could be extended to mango price prediction.…”

Section: Related Literaturementioning

confidence: 99%

Characteristic mango price forecasting using combined deep-learning optimization model

Jin²,

Huang³

et al. 2023

PLoS ONE

View full text Add to dashboard Cite

Accurate product price forecasting is helpful for scientific decision-making and precise industrial planning. As a characteristic fruit that drives regional development, mango price prediction is of great significance to several economies. However, owing to the strong volatility of mango prices, forecasting is vulnerable to uncertainties and is very challenging. In this study, a deep-learning combination forecasting model based on a back-propagation (BP) long short-term memory (LSTM) neural network is proposed. Using daily mango price data from a large fruit wholesale trading center in China from January 2nd, 2014, to April 18th, 2022, mango price changes are learned and predicted to support the fruit industry. The results show that the root mean-square error, mean absolute percentage error, and the R2 determination coefficient of the BP-LSTM combination model are 0.0175, 0.14%, and 0.9998, respectively. The prediction results of the combined model are better than those of the separate BP and LSTM models. Furthermore, it best fits the actual price profile and has better generalizability.

show abstract

“…Literature (Panigrahi & Behera, 2017; Purohit et al, 2021; Wang et al, 2013; Zhang, 2003) revealed the application of additive and multiplicative hybrid models employing statistical models along with ML models boost the forecasting performance of the resulting hybrid model. The better performance of these hybrid models is because of some interesting factors such as (a) most of the time series, including AQI time series, have linear patterns along with nonlinear patterns (Yin et al, 2021), (b) linear models cannot handle nonlinear patterns equally well, and nonlinear models cannot handle linear patterns equally well (Zhang, 2003), (c) the parallel application of linear and nonlinear models combine the advantages of individual models and hence provide greater capability to handle time series containing a combination of linear patterns and nonlinear patterns (Panigrahi et al, 2021; Purohit et al, 2021; Zhang, 2003), (d) nowadays it is computationally viable to execute multiple models concurrently and integrate the outputs of individual models to attain better performance Motivated by this, in this paper, we have proposed an Additive‐ARFIMA‐SVM model with functionally expanded (Sine, Cosine, and Square) inputs along with the original inputs for forecasting AQI. In our proposed model, we have chosen ARFIMA instead of ARIMA because ARFIMA is a fractional order signal processing technique that has superior capability than other integer‐order models like ARIMA in capturing short term as well as long term dependencies (Liu et al, 2017).…”

Section: Proposed Additive‐arfima‐svm Hybrid Model With Functionally ...mentioning

confidence: 99%

Study and development of hybrid and ensemble forecasting models for air quality index forecasting

Pradhan,

Panigrahi,

Purohit

et al. 2023

Expert Systems

Self Cite

View full text Add to dashboard Cite

In this paper, a viable, robust, and highly accurate additive hybrid model employing autoregressive fractionally integrated moving average (ARFIMA) and support vector machine (SVM) with functionally expanded inputs (Additive‐ARFIMA‐SVM) is presented for forecasting the air quality index (AQI). Additionally, thirteen additive and multiplicative hybrid models are introduced. Several alternatives in feature engineering employing functional expansion of inputs are incorporated to boost the performance of hybrid models. Furthermore, a gradient whale optimization algorithm with group best leader strategy (GWOA‐GBL) based meta‐heuristic algorithm is proposed. The missing values are imputed and a variable weight ensemble forecasting model is developed using the proposed GWOA‐GBL algorithm. To evaluate the effectiveness of the proposed Additive‐ARFIMA‐SVM forecasting model with functionally expanded inputs, comparisons are made with sixteen machine learning models, including long short‐term memory (LSTM), five statistical models, seventeen hybrid models, and ten variable weight ensemble models. Extensive statistical analyses are carried out on the obtained results considering four accuracy measures that show the statistical supremacy of the proposed Additive‐ARFIMA‐SVM model and GWOA‐GBL algorithm in predicting the AQI time series. The proposed Additive‐ARFIMA‐SVM model with functionally expanded inputs improves the AQI forecasting performance by 16.34% than autoregressive integrated moving average, 14.47% than ARFIMA, 33.96% than XGBoost, 43.47% than SVM, 49.39% than LSTM, 8.64% than Multiplicative‐ARIMA‐SVM model considering symmetric mean absolute percentage error. The proposed Additive‐ARFIMA‐SVM model is so efficient and reliable that it can be applied to forecast other time series like stock price, electricity load, crude oil price, sunspot number, stream flow, flood, drought etc.

show abstract

Time Series Forecasting of Price of Agricultural Products Using Hybrid Methods

Cited by 46 publications

References 8 publications

Estimation of Farmgate Rice Price Using Temporal Causal Modeling and Kalman Filtering

Estimation of Farmgate Rice Price Using Temporal Causal Modeling and Kalman Filtering

Characteristic mango price forecasting using combined deep-learning optimization model

Study and development of hybrid and ensemble forecasting models for air quality index forecasting

Contact Info

Product

Resources

About