MODELING AND FORECASTING OF TEA PRODUCTION IN INDIA
H.K.Niranjan1,B. Kumari2,Y. S. Raghav3,P. Mishra*, 4,A.M. G. Al Khatib5, M.Abotaleb6and Supriya7
1Agro Economic Research Centre ,JNKVV, Jabalpur-482004, (M.P.)
2 Department of Agricultural Economics, Rashtriya Kisan (PG) College (affiliated to Chaudhary Charan Singh University, Meerut), Shamli, India
3Departments of Mathematics, Faculty of Science, Jazan University, KSA.
4College of Agriculture, Powarkheda,J.N.K.V.V. (M.P.) India
5Departments of Banking and Insurance, Faculty of Economics, Damascus University, Syria
6Department of system programming, South Ural State University, Chelyabinsk, Russia
7Agricultural Economics, ANDUAT, Kumarganj, Ayodhya (U.P), India
*Corresponding author’s email: pradeepjnkvv@gmail.com
ABSTRACT
There are many measures of the importance of a crop to the economy, including its area, output, and yield increase. The current study will look at the growth rates of tea acreage, output, and yield in India using training data from 1918 to 2015 and testing data from 2016 to 2018. Using the data acquired, the ARIMA model and State Space Models were used to anticipate the area, production, and yield of tea from 2021 to 2027. According to the data, tea production in India is expected to reach 607 thousand hectares by2027, reflecting a 3.93 percent increase between 2021 and 2027. India's tea production is expected to reach 1486 thousand tonnes in2027, reflecting a 10.56 percent increase between 2021 and 2027. However, the tea production in India is expected to reach 2449 kg/hectare between 2021 and2027, reflecting a 4.12% increase over the preceding five-year period. The most essential tools for increasing tea production were area expansion and yield increase. As a result, the emphasis should be on expanding the area by exploiting available land and boosting productivity through technological advancements, varietal research, and the enhancement of agricultural advisory services across India.
Keywords: Growth rate, Trend, Area, ProductionTea, India, ARIMA, Forecast, State Space Models.
https://doi.org/10.36899/JAPS.2022.6.0569 Published first online June 11, 2022
INTRODUCTION
A close second to water in terms of global consumption is tea, with roughly two-thirds of the global population partaking (Khan and Mukhtar, 2013). Tea has a wide variety of products, from teabags to tea leaves, flavorings to non-flavorings, black tea to green tea, etc., allowing consumers to adjust to their own preferences and tastes (Karwowska, 2019). If the production process and post-harvest treatment are followed correctly, tea might be black, green, or oblong in color (Priya et al., 2015). In terms of production and consumption, black tea is the most popular in the world, followed by green tea, which is most popular in China and Japan, respectively (Wierzejska, 2014). Besides its wonderful flavor and perfume, tea offers a long list of health benefits, including the capacity to slow the growth of tumors, prevent heart disease and obesity, and lessen the risk of developing atherosclerosis (Wang et al., 2010). In addition to its numerous health benefits, tea is an important commercial crop in China, helping to drive the country's economy forward. In addition, it employs people from lower socioeconomic groups, such as women (Jain, 2011).
China, India, Sri Lanka, Keyna, and Indonesia are the world's top tea producers, together accounting for 70% of global production (Majumdar and Srirangam,, 2010; Zakir, 2017). Tea is a 200-year-old agro-based sector in India under the control of the country's Union government. In 1953, the Tea Act was passed, establishing the Tea Board to keep tabs on and promote the country's tea production, processing, exports, and imports. India is the world's second-largest tea producer, with the second-largest tea plantation area, and the fourth-largest tea exporter. India is the world's largest producer and exporter of tea, occupying around 12% of the global cultivable land and producing 20% of the world's total output. Although India is ranked 10th in the world in terms of tea production, its value yield is over double the world average (Basu et. al., 2010). While the plantation is indigenous to eastern and northern India, it is now grown in both the north and south of the country (Arya, 2013). North India contributes 86% of the country's overall output; the balance comes from the southern states of Tamil Nadu and Kerala. About half of India's tea is produced in Assam, which is the nation's leading tea producer. Because huge plantations aren't interested in expanding their land, small-scale tea planting is more feasible and practical, which is why it hasn't had as much success in non-traditional places (Kumar et. al., 2008).
Tea production forecasting is critical since the beverage crop is a key component of India's export basket, helping the country gain foreign cash. As a result, the current study is being undertaken to forecast tea production in the country. The ARIMA (Autoregressive Integrated Moving Average) model was chosen for prediction since it has been utilized and accepted by many other researchers. Mishra et al. (2012) used the ARIMA model to anticipate tea production, yield, and exports in India for the year 2020. Dhekale et al. (2014) used the ARIMA model to model and forecast tea output in West Bengal. They also created a new methodology for calculating weather and other indices that used direct influences from path coefficients and proved to be superior to the old methods. Rahman (2017) tested eleven ARIMA models and discovered that ARIMA (1,1,2) was the best fit for forecasting tea production in Bangladesh. Hossain and Abdulla (2015) also conducted tea forecasting in Bangladesh and discovered the ARIMA (1,2,1) model to be the best fit by utilizing the Run test and Jarque and Bera test analysis, which was accompanied by residual analysis.Mishra et al. (2021a) forecasted milk production in South Asian countries and China and found that all methods are not equally effective in forecasting. Mishra et al. (2021b) also carried out a study to find out the most suitable model for forecasting of sugarcane production in India.
MATERIALS AND METHODS
Dataset is yearly data from 1918 to 2020 about Area, Production and Yield Rate of Tea in India, we used training data from 1918 to 2015 and testing last 5 years from 2016 to 2020.Present study data is collected from www.indiastat.com.The flow diagram of present work given in figure A
Estimation Models: To forecast tea production up to year 2027 and used ARIMA model:
ARIMA Model:ARIMA models are the most commonly used statistical models for time series forecasting because they describe the autocorrelation in the data (Box et al., 2015). According to their nomenclature, these models are grouped into three categories (Autoregressive – Integrated – Moving Average) (p, d, q).
The term autoregressive (p) refers to the prediction of a variable using a linear set of its preceding values; the model of order p can be stated as (Mishra et al., 2021a):
Where: parameters of model, : lag order of the autoregressive process, : error term. Integrated (d) refers to the degree of stationary of a variable that is determined using ADF test (Devi et.al.,2021;Sarder et.al.,2021 and Mishra et.al.,2021c).
Moving average (q): uses past forecast errors in regression. The equation will be in the form:
Where: parameters of model, : lag order of the moving average, : error term.
Whereas (d) is decided by the ADF test, (p) and (q) are determined by the autocorrelation function (p) and the partial autocorrelation function R(p), which are supplied by Mishra et al., 2021b):
Figure A: Schema of Tea production in India
Figure 1: Actual and forecast values for Area, Production and Yield of tea in India with (Level -Total Signal-Slope -Autoregressive- Trend-Residuals) during the period 1937-2027 using State Space Models.
RESULTS AND DISCUSSION
The dataset contains annual data from 1918 to 2020 about the area, production, and yield rate of tea in India. We used training data from 1918 to 2015 and testing data from 2016 to 2020 for the last five years of the dataset.
Table 1: Summary Statistics, using the observations 1918 - 2020
Variable
|
Mean
|
Median
|
Minimum
|
Maximum
|
Area
|
390.21
|
353.36
|
237.57
|
579.35
|
Production
|
528.19
|
402.49
|
124.41
|
1377
|
Yield
|
1240.5
|
1146
|
434
|
2341
|
Variable
|
Std. Dev.
|
C.V.
|
Skewness
|
Ex. kurtosis
|
Area
|
95.78
|
0.24
|
0.90
|
-0.52
|
Production
|
344.30
|
0.65184
|
0.80636
|
-0.42193
|
Yield
|
533.72
|
0.43024
|
0.23555
|
-1.1196
|
From 1918 to 2020, India's tea acreage expanded from (237.57) thousand to (579.35) thousand hectares, according to our research. There were (390.21) thousand hectares of tea plantations in India on average. Small outliers were evident in the data, as seen by an exaggerated value of Kurtosis (-0.52). Skewness (0.90) indicates that there is a prospect of growth in India's tea region. There has been a significant increase in the amount of tea produced in India throughout this time period. The country's annual tea production was (528.19) metric tons on average. Example Kurtosis was (-0.42), which indicates a mesokurtic distribution of the data. The presence of a positive skewness (0.80) value suggests that India's tea production could expand in the future. From (434) kilograms per hectare to (2341) kilograms per hectare, the tea yield in India has risen throughout this time span. India's average yield of (1240.5) Kg. / Hectare of tea. Example: The (-1.11) kurtosis number indicates that there are some minor anomalies in the data. The distribution is nearly symmetric after a positive skewness (0.23) of between -0.5 and 0.5.
Table 2: ARIMA Model fitted for Area, Production and Yield of tea in India for training data set (1918 to 2015)
|
MODEL
|
Akaike Information Criterion (AIC)
|
Bayesian Information Criterion (BIC)
|
MAE
|
RMSE
|
MAPE
|
Area
|
ARIMA
(0,1,1)
|
833.5
|
841.41
|
6.72
|
13.97
|
1.87
|
Production
|
ARIMA
(4,1,4)
|
960
|
986.32
|
17.15
|
23.13
|
4.25
|
Yield
|
ARIMA
(0,1,1)
|
1153.9
|
1161.79
|
47.83
|
67.2
|
4.84
|
Table 2 lists the top selected ARIMA models, which were chosen based on the following criteria: Akaike,Root mean squared error, Mean absolute error, Mean absolute percentage error, Maximum number ofsignificantcoefficients.
Table 3: State Space Models for Area, Production and Yield of tea in India for training data set (1918 to 2015)
|
Structural time series models for Area, Production and Yield of tea in India
|
|
|
|
Component
|
Value
|
Std.Err
|
t-stat
|
Prob
|
Akaike Information Criterion (AIC)
|
Bayesian Information Criterion (BIC)
|
within-sample RMS
|
MAE
|
MAPE
|
Area
|
Level
|
569.5218
|
8.832
|
64.4873
|
0
|
8.1891
|
8.3738
|
6.7227
|
2.9889
|
0.861
|
Slope
|
4.5018
|
1.996
|
2.2549
|
0.0264
|
AR1
|
0.0807
|
5.393
|
0.015
|
0.9881
|
Production
|
Level
|
1219.87
|
36.097
|
33.7942
|
0
|
9.1392
|
9.3766
|
2.8436
|
2.1367
|
0.5759
|
Slope
|
23.97
|
3.886
|
6.1676
|
1.83E-08
|
AR3
|
-11.28
|
35.566
|
0.3172-
|
0.7518
|
Yield
|
Level
|
2086.02
|
94.674
|
22.0337
|
0
|
11.2348
|
11.393
|
20.9348
|
14.7584
|
1.5794
|
Slope
|
15.8
|
3.983
|
3.9668
|
1.41E-04
|
AR1
|
45.47
|
92.267
|
0.4928
|
0.6233
|
Table 3 provides us with an estimate of the parameters for the State Space Models (SSMs). As a result of these factors, the training data set ("1918 to 2015") has the best model based on the lowest possible AIC/BIC/RMSE/MAE/MAPE values. When it comes to predicting tea production, the state space model beats ARIMA. In terms of prediction accuracy, the state space model has far higher accuracy than the ARIMA model, because all the accuracy metrics (AIC, BIC, RMSE, MAE and MAPE) are lower than the ARIMA model's. It is not necessary to require stationarity in these models for the state space model to reflect the complicated non-linear character of data series with a variety of distinct specifications, structural breaks, shifts, time-varying factors, and missing data. Dynamic timeseries can also be modeled using the state space paradigm. The ARIMA model must be stationary, while models that incorporate unobserved components may need to be employed to eliminate trend and seasonal effects from the ARIMA model. RMSE, MAE, and MAPE values in Table 4 show us the best State Space Models based on the testing data set ("2016 to 2020").
Table 4: RMSE, MAE and MAPE for testing data set (2016 to 2020)
|
RMSE
|
MAE
|
MAPE
|
Area
|
5.33
|
5.12
|
0.8863
|
Production
|
52.94
|
38.34
|
2.9435
|
Yield
|
50.14
|
38.42
|
1.6925
|
Table 5: Forecasting from 2021 to 2027 for Area, Production and Yield of tea in India using best models
Year
|
Area
|
Production
|
Yield
|
2021
|
584
|
1344
|
2352
|
2022
|
587
|
1358
|
2365
|
2023
|
592
|
1375
|
2379
|
2024
|
595
|
1424
|
2392
|
2025
|
599
|
1422
|
2406
|
2026
|
603
|
1462
|
2420
|
2027
|
607
|
1486
|
2449
|
As shown in Table 5, the tea area in India is predicted to grow by 2027 to reach 607 thousand Hectares (about). During the decade 2021-2027, the economy will increase at a rate of 3.93 percent. The output of tea in India is predicted to reach 1486 thousand tonnes in 2027, representing a growth rate of 10.56 percent between 2021 and 2027. During the period 2021-2027, the tea yield in India is predicted to reach 2449 Kg. Hectare, representing a growth rate of 4.12 percent over the previous five-year period.
Forecasting:Following the development of the top models, forecasting for the area, production, and yield of tea in India is carried out. With respect to all time series, the residuals of the chosen models were determined to be stationary and white noise. Using the best-fitted models, the anticipated values for the years "2021" to "2027" are depicted in the figures. The figures displayed the anticipated values; all forecasted lines in the figures are within a few percentage points of the actual values, demonstrating the usefulness of the models chosen. From the forecast numbers, it can be seen that the tea growing area, tea production, and tea yield are all expected to increase in the future.
Conclusion:During the last 102 years (from 1918 to 2020), tea production has increased significantly in both land area and yield. Since 1918–1981, the rate of yield rise was much larger than in the period 1982–2020, whereas the rate of area expansion had the reverse tendency. Both when technology was more conventional and when it began to improve in 1918–2020, considerable output expansion was a result of increased yields, notably in that time period. With the expansion of "small-scale gardening" as an approach to tea cultivation, the industry's output is rising, bringing prosperity to many farmers and creating work for many people — primarily women. The long-term strategy for increasing tea production in the northern areas of India needs to be improved. Thus, India can continue to expand the area where tea can be grown as long as the region permits, but ultimately, she will have to rely on a strategy to boost yield. In order to take use of these possibilities, the administration of the Bangladesh tea plantations should look for suitable extra areas and put them under tea production. Redouble efforts to improve the tea plantation's yields and quality by bringing in higher yielding and better-quality vegetative clones and seeds from BTRI.
Acknowledgments: The work was supported by Act 211 Government of the Russian Federation, contract No. 02.A03.21.0011. The work was supported by the Ministry of Science and Higher Education of the Russian Federation (government order FENU-2020-0022).
REFERENCES
- Arya, N. (2013). Indian tea scenario. International Journal of Scientific and Research Publications, 3 (7): 1-10.
- Basu, M.A., Bara, B. and Rajan, A. (2010) Tea statistics: global scenario. Journal of Tea Science. 8(1): 121-124.
- Dhekale, B. S., P. K. Sahu, K. P. Vishwajith, P. Mishra and M. D. Noman (2014) Modeling and forecasting of tea production in West Bengal. Journal of Crop and Weed, 10(2) : 94-103.
- Devi, M., J. Kumar, D. P. Malik and P. Mishra (2021). Forecasting of wheat production in Haryana using hybrid time series model. Agri. Food Res., :100175.
- Hossain, M. M. and F. Abdulla (2015) Forecasting the tea production of Bangladesh: application of arima model. Jordan J. Math. & Stat., 8(3): 257-270.
- Jain, S. (2011) Indian tea industry- an analysis using concept of forward engineering, ABES-IT Group of Institutions.
- Karwowska, K., M. Skotnicka and M. Smiechowska (2019) Tea production and its forecasts and the possibility of tea cultivation in the context of environmental requirements in China. Scientific J. Warsaw University of Life Sci., 19 (1): 180-191.
- Kumar, P., P. S. Badal, N. P. Singh and R. P. Singh. (2008) Tea industry in India: problems and prospects. Indian J. Agri. 63 (1): 84-96.
- Khan, N., and H. Mukhtar. (2013) Modulation of signaling pathways in prostate cancer by green tea polyphenols. Biochemical pharmacology, 85(5): 667-672.
- Mishra, P., P. K. Sahu, P. Bajpai and H. K. Niranjan (2012) Past trends and future prospects in production, and export scenario of tea in India. Intl. Review Business and Finance, 4(1): 25-33.
- Mishra, P., A. Matuka, M. S. A. Abotaleb, W. P. M. C. N. Weerasinghe, K. Karakaya and S. S. Das (2021a) Modeling and forecasting of milk production in the SAARC countries and China. Modeling Earth Systems and Environment, 1-13.
- Mishra, P., A. M. G. Al Khatib, I. Sardar, J. Mohammed, K. Karakaya, A. Dash and A. Dubey (2021b) Modeling and forecasting of sugarcane production in India. Sugar Tech, 23(6): 1317-1324.
- Mishra P., A. Yonar. H. Yonar, B. Kumari, M. Abotaleb, S. S. Das and S. G. Patil (2021c), State of the art in total pulse production in major states of India using ARIMA techniques. Current Research and Food Science, 4:800–806. doi. :10.1016/j.crfs.2021.10.009.
- Majumdar, S., and Srirangam (2010). Potential of the bioflavonoids in the prevention/treatment of ocular disorders. J. Pharm. and Pharmacology, 62(8): 951-965.
- Priya, B. M., V. Anitha, M. Shanmugam, B. Ashwath, S. D. Sylva and S. K. Vigneshwari (2015). Efficacy of chlorhexidine and green tea mouthwashes in the management of dental plaque-induced gingivitis: A comparative clinical study. Contemporary Clinical Dentistry, 6(4): 505-506.
- Rahman, A. (2017) Modeling of tea production in Bangladesh using autoregressive integrated moving average (ARIMA) model. J. Applied & Computational Math., 6(2): 349-351.
- Sardar, I., S. Aflake, K. Karakaya, T. Makarovskikh, M. Abotaleb and P. Mishra (2021). Machine learning-based Covid-19 forecasting: Impact on Pakistan stock exchange. Intl. J. Agri. & Stat. Sci., 17(1): 53-61.
- Wang, H., Y. Wen, Y. Du, X. Yan, Guo, J. A. Rycroft, N. Boon, Eva M. R. Kovacs and D. J. Mela (2010) Effects of catechin enriched green tea on body composition. Obesity, 18: 773-779.
- Wierzejska, R. (2014). Tea and health–a review of the current state of knowledge. Przegl Epidemiol, 68(3), 501-506.
- Zakir, M. (2017) Review on tea (Camellia sinensis) research achievements, challenges and future prospective including Ethiopian status. Intl. J. Forestry and Hort., 3 (4): 27-39.
|