Forecasting pearl millet area and production in India using regression models

Introduction

Bajra, or pearl millet (Pennisetum glaucum), is a grain crop that is produced mostly in Asia and Africa in tropical semi-arid regions of the world. Bajra can be cultivated in regions where other cereal crops, like wheat or maize, would not thrive since it is well suited to production systems with little rainfall (200-600 mm), low soil fertility, and high temperatures. It is an important fact that our country, India, is termed as the original home of nutria cereals (shree Anna millet). Pearl millet requires minimal fertilizer and pesticide inputs and is resistant to drought and other harsh weather. India is the world's top producer of pearl millet, accounting for over 44% of global production. Nearly 51% of India's gross cultivated area is made up of pearl millet. In India, pearl millet crops were grown on 7.41 million hectares of land in 2019-20. With almost 28 million tons of pearl millet produced, or 51% of India's total production, Rajasthan leads the country in this regard. The world's production of pearl millet has increased over the past 15 years mostly due to the availability of improved genetics, the adoption of hybrids in India, and the extension of the area under pearl millet cultivation in West Africa. The production of pearl millet is hampered by a number of biotic and abiotic stressors, which significantly reduce yields. India is the world's greatest producer of pearl millet, covering 7.14 million hectares and producing 8.06 million tons with an average yield of 1132 kg/ha (http://www.aicpmip.res.in/pmnews 2017. pdf) and continues to play a critical role in food and livelihood security of millions of people in India. It is the main source of calorie intake for the 500 million people of Asia and Africa. It has a brief life cycle, tiny seeds, and a C4 panicoid, highly cross-pollinated annual crop. Pearl millet, a warm-season coarse grain, may flourish in barren soils with little water-holding capacity, limited rainfall, high temperatures, and droughty conditions when other cereals would not. Food, feed, fodder forage, brewing, biofuel, building materials, and fuel for cooking in dryland environments are just a few of the many use for this crop. In data science, regression is a crucial technique for predicting a range of continuous targets, which helps to prevent this work from being overly demanding and from being seen as comprehensive in the actual world. Regression-related studies are still thoroughly assessed by scholars today since they receive the most attention. The area and production of pearl millet were fitted and forecasted in the current study using regression modeling.

Regression Study

Regression analysis uses data analysis to determine the link between response variables and explanatory variables. The variables chosen for examination in this study are production, area, and year. For this study, year is acknowledged as an explanatory variable, while pearl millet production and area are regarded as response variables.

The different steps undertaken for conducting this study is illustrated as a flow chart in Fig. 2. Data on pearl millet production and area was gathered from Agricultural Statistics at a Glance 2022, Government of India, during a 71-year period (1951-2021). Simple regression analysis was used to model the area and production of pearl millet. The association between an independent and a dependent variable was used to forecast the response variable's future observation. As a result, the study was carried out with the presumption that the projected future values for pearl millet production and area would continue the previous pattern. The primary advantage of this strategy was described as its simplicity and capacity to forecast any available data. Acceptance of the fitted model and the model's advanced independent variable requires, F test was performed. The fitted model's significance was determined using the F test. The model is accepted or rejected depending on whether the F value if the F value obtained after undergoing F test is higher than the F table value (from F distribution table), the model is accepted or otherwise rejected. By examining the performance of regression models developed for the data, the optimal forecasting model was found. The chosen criterion for verifying was the lowest value for RMSE and MAPE. Fitted models' work (Eqs. 7

Modeling the production and area of pearl millet

The regression analysis's findings using various techniques are explained in Table 2. The 2 R valuesa lie above 0.969 and0.870 for the fitted regression models, shown a significant degree of correlation between area and years, production and year, respectively. Additionally, the determination coefficients showed that only a small percentage (0.011-0.010%) of the variance in area and production may be attributed to factors other than the predictors (see Eqns. 1to 6). The cubic model is the most suitable for predicting the area and production of pearl millet, according to the model evaluation criteria (Eqns. 9 and 10). The above describes the fitted models' validation outcomes. The calculated value of F is higher than the F table which confirmed that Eqns. ( 7) to ( 8) are significant at 95% confidence level. Therefore, it was determined that the explanatory variables used in this study to create the regression models for pearl millet production and area are significant.

Forecasting Pearl Millet

The cubic model built for area and production yielded MAPE values of 0.053 and 0.162, respectively (see Table 2, 3). These results showed that pearl millet production and area could be effectively estimated using the cubic model, which was selected as the best forming model. The MAPE values of the cubic models showed that they are both highly accurate prediction models for the area and production of pearl millet (see Table 2, 3). Furthermore, it is clear from Fig. 3 that the actual and expected values for the area and output of pearl millet are very close to each other. In order to model the area and production of pearl millet, the researchers will need to apply other models such as artificial intelligence, grey theory forecasting, hybrid approach, multiple linear regression, and other econometric methods in the future. Future research findings can be evaluated by comparing them to the current study.

References

1. USDA Foreign Agricultural Service. Crop Explorer: Commodity View (Millet). Available from: https://ipad.fas.usda.gov/cropexplorer/cropview/commodi tyView.aspx?cropid=0459100&sel_year=2018&rankby= Production
2. Altman N, Krzywinski M. Crafting 10 years of statistics explanations: Points of significance. Annu Rev Stat Appl. 2025;12(1):69-87.
3. India. Ministry of Agriculture and Farmers Welfare. Department of Agriculture. Agricultural statistics at a glance 2022. New Delhi: Government of India; 2022. p.72.
4. Chicco D, Warrens MJ, Jurman G. The coefficient of determination R-squared is more informative than SMAPE, MAE, MAPE, MSE and RMSE in regression analysis evaluation. PeerJ Comput Sci. 2021;7:e623.
5. Yadav RS, Mehta V, Tiwari A. An application of time series ARIMA forecasting model for predicting nutri cereals area in India. Pharma Innov J. 2022;11(3):12601267.
6. Yadav RS, Mehta V, Krishna G. Regression models for forecasting nutri cereals area and production in India. Pharma Innov J. 2023;SP-11(3):1260-1267.
7. Dhankar N, Routroy S, Sharma SK. A multiple regression model to analyze the impact of endogenous factors on pearl millet and stover yield in India. J Model Manag. 2026;21(2):584-612.
8. Kutty VK. Regression model. In: Global millets production for a sustainable future. 2025. p.231.