You are on page 1of 19

STARTING WITH THE NAME OF ALLAH, THE

MOST GRACIOUS, THE MOST MERCIFUL.


TIME SERIES MODELING FOR FORECASTING
MAIZE PRODUCTION OF PAKISTAN

• By
• Ahmad Jawad
• BSTF11E091
• BS (Hons) in Statistics
• Supervised by
• Sir Qaisar Rashid
MAIZE
• Maize, commonly known as corn, is a large grain plant first domesticated by indigenous
peoples in Mexico about 10,000 years ago. The six major types of corn are dent corn, flint
corn, pod corn, popcorn, flour corn, and sweet corn.
MAIZE PRODUCING COUNTRIES
• It is the 3rd biggest yielding crop of the world.
• America is the largest producer of maize with the numbers of
345,486(1000 MT) China is on 2 nd place with 224580(1000 MT)
and Brazil on 3rd with 81500(1000 MT). Pakistan stands at 21
positions with 5100(1000 MT).
MAIZE PRODUCTION IN PAKISTAN OVERVIEW
• In Pakistan maize is one of the biggest crop cultivated.4.8% of
the cultivated area is covered with maize, 0.9 million hectare
area is under maize cultivation. Yearly production of maize is 1.3
million tons.97% maize is produced in two provinces Kpk and
Punjab. Area wise Kpk covers 57% of the total area and
production wise it covers 68% of total production, Punjab covers
38% area wise and 30 % production wise, Baloshistan and
Sindh only take part in 2-3 % production of maize.
SOME USES OF MAIZE
• Maize has a lot of uses maize flour is used to make breads and a thickening
agent for custards and jellies.
• Popcorn a world famous snack is produced when maize is heated at a
certain temperature.
• Starch which is extracted from maize is used for making noodles.
• The corn syrup of maize has high fructose and works as a sweetener, it also
keeps certain foods moist.
• Fabrics and Plastics are made from maize stocks.
• Maize is used to feed life stock and as forage after the process of
fermentation of corn stocks.
OBJECTIVE
• Since Maize is one of the main crops of Pakistan so it is important to keep a track of
its production. For the future growth of maize, forecasting is the main tool for
predicting its production. Forecasting is useful for country planning.
TIME PLOT FOR MAIZE PRODUCTION IN
PAKISTAN
METHODOLOGY
• Required time series data for this research project was collected from a
famous website called www.indexmundi.com and a government organization
which is known as Federal Bureau of Statistics. In time series there are
various models available for forecasting, in this project i will use Box Jenkins
Methodology (1976) to estimate the future challenges in advance. The data
selected is first checked in a software called e views. To check stationarity
ADF test is applied with α taken as 0.01, 0.05 and 0.10. the data came
stationary at level so i selected ARIMA (p,d,q) model and made its
correlogram. P is the order of autoregressive model, d is the difference and q
is the order of moving average model. After finding out the values of ACF
and PACF from correlogram i made pairs of models.
Market Year Production Unit of Measure Growth Rate Market Year2 Production3 Unit of Measure4 Growth Rate5
1960 439 (1000 MT) NA 1988 1204 (1000 MT) 6.83%
1961 488 (1000 MT) 11.16% 1989 1179 (1000 MT) -2.08%
1962 489 (1000 MT) 0.20% 1990 1185 (1000 MT) 0.51%
1963 526 (1000 MT) 7.57% 1991 1100 (1000 MT) -7.17%
1964 528 (1000 MT) 0.38% 1992 1100 (1000 MT) 0.00%
1965 540 (1000 MT) 2.27% 1993 1213 (1000 MT) 10.27%
1966 587 (1000 MT) 8.70% 1994 1318 (1000 MT) 8.66%
1967 792 (1000 MT) 34.92% 1995 1504 (1000 MT) 14.11%
1968 626 (1000 MT) -20.96% 1996 1491 (1000 MT) -0.86%
1969 668 (1000 MT) 6.71% 1997 1517 (1000 MT) 1.74%
1970 717 (1000 MT) 7.34% 1998 1665 (1000 MT) 9.76%
1971 705 (1000 MT) -1.67% 1999 1652 (1000 MT) -0.78%
1972 715 (1000 MT) 1.42% 2000 1643 (1000 MT) -0.54%
1973 767 (1000 MT) 7.27% 2001 1664 (1000 MT) 1.28%
1974 747 (1000 MT) -2.61% 2002 1737 (1000 MT) 4.39%
1975 803 (1000 MT) 7.50% 2003 1897 (1000 MT) 9.21%
1976 764 (1000 MT) -4.86% 2004 2797 (1000 MT) 47.44%
1977 821 (1000 MT) 7.46% 2005 3109 (1000 MT) 11.15%
1978 798 (1000 MT) -2.80% 2006 3088 (1000 MT) -0.68%
1979 875 (1000 MT) 9.65% 2007 3605 (1000 MT) 16.74%
1980 946 (1000 MT) 8.11% 2008 3593 (1000 MT) -0.33%
1981 930 (1000 MT) -1.69% 2009 3262 (1000 MT) -9.21%
1982 1005 (1000 MT) 8.06% 2010 3707 (1000 MT) 13.64%
1983 1013 (1000 MT) 0.80% 2011 4338 (1000 MT) 17.02%
1984 1028 (1000 MT) 1.48% 2012 4220 (1000 MT) -2.72%
1985 1009 (1000 MT) -1.85% 2013 4944 (1000 MT) 17.16%
1986 1111 (1000 MT) 10.11% 2014 4695 (1000 MT) -5.04%
1987 1127 (1000 MT) 1.44% 2015 5100 (1000 MT) 8.63%
ADF TEST AT FIRST DIFFERENCE
Augmented Dickey-Fuller Test Equation
Dependent Variable: D(Y,2)
Method: Least Squares
Date: 05/03/16 Time: 22:50
Sample(adjusted): 1963 2015
Included observations: 53 after adjusting endpoints
Variable Coefficient Std. Error t-Statistic Prob.
D(Y(-1)) -1.002332 0.211254 -4.744676 0.0000
D(Y(-1),2) -0.100519 0.146421 -0.686508 0.4956
C 86.61994 34.95181 2.478268 0.0166
R-squared 0.548930 Mean dependent var 7.622642
Adjusted R-squared 0.530887 S.D. dependent var 322.2945
S.E. of regression 220.7454 Akaike info criterion 13.68684
Sum squared resid 2436426. Schwarz criterion 13.79836
Log likelihood -359.7011 F-statistic 30.42374
Durbin-Watson stat 2.030887 Prob(F-statistic) 0.000000
CORRELOGRAM
PAIRS
Pairs AIC-values SBC-values

(0,1,0) 13.59730 13.63380

(0,1,1) 13.62459 13.69758

(1,1,0) 13.64157 13.71524

(1,1,1) 13.67721 13.78770

(2,1,0) 13.65938 13.73374

(2,1,1) 13.67721 13.78770


SELECTION OF MODELS THROUGH AIC AND SBC

Pairs AIC-values SBC-values

(0,1,0) 13.59730 13.63380

(0,1,1) 13.62459 13.69758

(1,1,0) 13.64157 13.71524


BEST FITTED MODEL
• Model (1,1,0)
FORECASTED VALUES

year production
2005 4221.112
2006 4305.876
2007 4390.64
2008 4475.405
2009 4560.169
2010 4644.933
2011 4729.697
2012 4814.461
2013 4899.225
2014 4983.989
2015 5068.754
2016 5153.518
2017 5238.282
2018 5323.046
2019 5407.81
2020 5492.574
TIME PLOT
CONCLUSION
• We know that population of Pakistan is increasing rapidly, so it is necessary to plan
the future requirements of the country. For this purpose, forecasting is used to find
out about the need of nation in advance. Maize is the basic need of any country all
over the world. In this study, i developed time series models to forecast maize
production of Pakistan on the basis of historical data i.e. 1960-2015. I have
developed different time series models on maize production of Pakistan on this data.
Best model is selected on the basis of model selection criteria which is AIC and
SBIC. The main cause of preferring time series model as other studies is that the
model best fitted has also satisfied all the assumptions of residuals which are,
independence, normality and no autocorrelation. On the basis of these model
selection criteria, i have found that best fitted model for maize production forecasting
of Pakistan is ARIMA (1,1,0) because this model has lower values of AIC and SBIC
as compared to other fitted models. On the basis of this model, i have found that
wheat production of Pakistan would become 5492.574 metric tons in 2020.

You might also like