Prediction of Poor Inhabitant Number Using Least Square and Moving Average Method

—The number of poor inhabitant in South Kalimantan decreased within the last three years compared with the previous years. The numbers of poor inhabitant differs from time to time. This scaled dynamical number has been a problem for the local government to take proper polices to solve this matter. It will then be necessary to predict a potential number of poor inhabitants in the next year as the basis on subsequent policy making. This research will apply both Least Square and Moving Average method as the measurement to count prediction values. From the results of the study, the prediction analysis by using those two methods is valid for predicting acquired number of poor inhabitant for the next period according to the data from the previous year. Based on the study, the validity of Least Square was 98.35% and Moving Average was 98.79% by using the data in the last seven years.


INTRODUCTION
Poverty is the main problem in South Kalimantan Province, Indonesia [1]. According to Statistic Center Agency, the poor inhabitant is defined as those who averagely spend below the poverty line per capita per month [2]. Based on the Statistic Center Agency in South Kalimantan [2], the number of poor inhabitant decreased each year. In 1999 the number of poor inhabitant was 440,200 and while at the end of 2014 became 182,876 inhabitants. From those data, the number of poor inhabitant decreased 5.28% on average each year and in the last three years decreased 3.31%. There are a decreasing numbers of poor inhabitants in South Kalimantan due to the numbers of building.
The problem was in predicting the number of poor inhabitant in South Kalimantan that decreased in the last three years. However, the prediction cannot be predicted for the next years, and it caused the local government had difficulty to make decision. This study was aimed to decide the way of predicting the number of poor inhabitant in South Kalimantan for the years to come by using Least Square and Moving Average method. Hopefully, the result of the research could help the government to increase the people's life quality.

II. RESEARCH METHOD
Time series analysis was a statistical analysis method applied to predict a future condition. To make an accurate data, the prediction was conducted for a very long time and much data was needed. As one of the choices to describe a future trend, time series analysis can be applied to reflect dynamic variable from one time to another [3]. From the previous studies by using Least Square [4][5][6][7][8] and Moving Average [3] [9][10][11][12][13][14] method, the data and analysis showed the future prediction. It was defined as a management process in making decision. It was described as a prediction process in the unknown future situation. In general term, it was well known as a prediction referring to time series estimation or longitudinal type of data [9].
The Least Square method was often used to predict (Y), due to its detail measurement [4]. The trend line (1) was: : Scaled data (time series) = Trend value prediction. a0 : Trend value in the basis year. b : Average growing trend value in each year. X : Time variable (year). To conduct the calculation, a certain value in time variable (X) was required so that the total variable score was zero or ΣX=0. In analyzing the data with Least Square method, it is generally divided into two parts i.e. "even data" and "odd data" [4].
For odd "n", where: The interval between two times was one-unit value It was marked as negative when it was above 0 It was marked as positive when it was below 0 For even "n", where: The interval between two times gains two-unit value It was marked as negative when it was above 0 It was marked as positive when it was below 0 Generally, linier line equation from time series analysis (2) was: a bX  Description: Y is a variable that trend was searched.
X is a time variable (year).
Meanwhile, to find constant value "a" and parameter value "b" (3) was: www.ijacsa.thesai.org a = ΣY / n, and b = ΣXY / Σ  Moving Average method was a prediction approach by taking some observed groups of the values, finding the average, and using the average values as a prediction of subsequent period. The formula (4) was [10]: : Forecast for the coming period n : Number of period to be averaged "n" : Actual Occurrences in the past period, two period ago, three period ago, and so on respectively.

A. Least Square Method
In this study, the data for the "odd data", previous collected data from the last nine years were required. Meanwhile, when processing data tabulation for an "even data", the previous data collection from the last ten years are required.  Thus, to find "a" value was: If the different score between Least Square method was >40%, it was considered to be invalid. Compared to the accurate score in 2008, the different was 17.99% (39,387 inhabitants) and the data was valid. In 2009, the different was 6.80% (11,960 inhabitants), it means that the data was valid as well. In 2010, the different was 1.65% (3,009 inhabitants), it means that the data was valid. In 2011 the difference was 6.70% (13,043 inhabitants), it is also means that the data was valid. In 2012, the comparison was 8.39% (15,988 inhabitants), the data also was valid. In 2013, the difference was 7.20% www.ijacsa.thesai.org (13,274 inhabitants), the data was considered to be valid. And in 2014, the comparison was 11.53% (21,086 inhabitants) it means that the data was valid as well. Based on the seven differences, the all data was valid. So, Least Square method was effective or accurate.
The next phase was calculating the prediction number of poor inhabitant in 2015. Based on the data tabulation for "odd data", the poor inhabitant data were needed from the last nine years, starting from 2006 to 2014. The table 3, showed that the "a" and "b" values were obtained. To count "a" and "b" values, the following formula was applied: To find out "a" value was:    Based on table 5, "a" and "b" values were obtained. To find those scores, the following formula was applied: To find the "a" value was: It means that the prediction number of poor inhabitant was 160,493 inhabitants.
From the calculation of Least Square method, the prediction number of poor inhabitant in 2015 for "odd data" was 157,254 inhabitants. And for the "even data" was 160,493 inhabitants. So, the different was 2.02%. However, the result of

B. Moving Average Method
Before measuring the prediction number of poor inhabitant in 2015, the test of number of poor inhabitant in 2008,2009,2010,2011,2012,2013, and 2014 by using Single Moving Average method were conducted to know whether the data was valid or not compared with the accurate data of poor inhabitant in the last seven years. To count the prediction number in 2015, the prediction number in 2014 was counted at first. If the different between prediction calculation with Single Moving Average for two periods with the results was >40%, then it is considered to be invalid. Compared to the accurate data in 2008, the different was 14.48% (37,078 inhabitants), it was considered to be valid. In 2009, the different was 22.20% (50,222 inhabitants), it means that the result was considered to be valid. In 2010, the different was 7.84% (15,475 inhabitants), it means that the data was valid. In 2011, the comparison was 8.04% (15,653 inhabitants), the data was valid. In 2012, the difference was 1.21% (2,304 inhabitants), the data was also valid. While in 2013, the difference was 4.32% (8,313 inhabitants) and the data was valid. Lastly, in 2014 the difference was 2.44% (4,571 inhabitants) it also means the data was valid. According to those seven comparisons, the use of Single Moving Average was effective.

C. The Comparison Result of Least Square and Moving
Average Method According to Least Square and Moving Average method, if the difference between prediction calculation with the result was >40% it means that the data were invalid. Based on table 8, there was a comparison result between Least Square and Moving Average method for the last seven years.  In accordance with the figure 3, the lower data for Least Square method was 1.65% and 1.21% for Moving Average method, so the data was considered to be effective. The validity of Least Square and Moving Average was based on the accurate measured data.

IV. CONCLUSION
The use of Least Square and Single Moving Average method was effective to predict the number of poor inhabitant in South Kalimantan for the next period. If the difference between the calculation of Least Square and the accurate result was >40%, it was considered to be invalid. Compared to the actual result in 2008, the difference was 17.99% (39,387 inhabitants), it means that the result was valid. In 2009, the difference was 6.80% (11,960 inhabitants), it means that the result was valid. In 2010, the comparison was 1.65% (3,009 inhabitants), it also considered to be valid. In 2011, the comparison was 6.70% (13,043 inhabitants) and the data was valid. In 2012, the difference was 8.39% (15,988 inhabitants) and the data was valid. In 2013, the comparison was 7.20% (13,274 inhabitants) it means that the data was valid. And in 2014, the difference was 11.53% (21,086 inhabitants) and the data was considered to be valid. It means that Least Square method was approximately effective. inhabitants. It was considered to be invalid when the difference between the calculation with Single Moving Average for two periods and the accurate result was >40%. Compared to the accurate data in 2008 the difference was 14.48% (37,078 inhabitants), it means that the result was valid. In 2009, the difference was 22.20% (50,222 inhabitants), and the data was valid. In 2010, the difference was 7.84% (15,475 inhabitants), it means that the result was also valid. In 2011, the comparison was 8.04% (15,653 inhabitants) it means that the data was valid. In 2012 the difference was 1.21% (2,304 inhabitants), and the data was considered to be valid. In 2013, the comparison was 4.32% (8,313 inhabitants), it also considered to be accurate. In 2014, the difference was 2.44% (4,571 inhabitants) it means that the data was valid. Based on the seven comparisons, the all data was accurate or valid. Thus, Single Moving Average was approximately effective.
The accurate result of Least Square was 98.35% and 98.79% for Moving Average, so it was considered to be valid in predicting the number of poor inhabitants.
For the next researches, the number of data and additional variable are required. Smart system can be used as a method to predict the number of poor inhabitant.