Calculation of ARIMA Comparison of forecasting results

38 Table 4.20 Error Calculation Of Moving Average Consumption Years Consumption Actual Data Forecast MA=4 Absolute deviation Forecast MA=5 Absolute deviation 2000 14,522,025.00 2001 15,208,543.00 2002 15,445,816.00 2003 16,702,230.00 2004 17,931,957.00 15,469,653.50 2,462,303.50 2005 19,909,502.00 16,322,136.50 3,587,365.50 15,962,114.20 3,947,387.80 2006 21,730,193.00 17,497,376.25 4,232,816.75 17,039,609.60 4,690,583.40 2007 23,016,749.00 19,068,470.50 3,948,278.50 18,343,939.60 4,672,809.40 2008 27,748,242.00 20,647,100.25 7,101,141.75 19,858,126.20 7,890,115.80 2009 29,427,305.00 23,101,171.50 6,326,133.50 22,067,328.60 7,359,976.40 2010 31,680,704.00 25,480,622.25 6,200,081.75 24,366,398.20 7,314,305.80 2011 27,968,250.00 26,720,638.60 MAD 2,796,825 2,672,064 Table 4.21 Error Calculation Of Moving Average Population Years Population Actual Data Forecast MA=4 Absolute deviation Forecast MA=5 Absolute deviation 2000 5,231,189.00 2001 5,331,311.00 2002 5,434,293.00 2003 5,541,062.00 2004 5,652,797.00 5,384,463.75 268,333.25 2005 5,769,709.00 5,489,865.75 279,843.25 5,438,130.40 331,578.60 2006 5,893,738.00 5,599,465.25 294,272.75 5,545,834.40 347,903.60 2007 6,023,053.00 5,714,326.50 308,726.50 5,658,319.80 364,733.20 2008 6,149,620.00 5,834,824.25 314,795.75 5,776,071.80 373,548.20 2009 6,262,667.00 5,959,030.00 303,637.00 5,897,783.40 364,883.60 2010 6,355,112.00 6,082,269.50 272,842.50 6,019,757.40 335,354.60 2011 6,197,613.00 6,136,838.00 MAD 619,761 613,684 The describ in detail to calculate the error of forecasting in 2022. Next, the calculate error forecasting uses Moving Average and make sure uses MAD.

4.3.12 Calculation of ARIMA

The first column is time or period. The second column is t period or time. The third column are actual data. 1. The first step input actual data 2. Find forecasting with averages then use formulation moving averages commit to user 39 MA =   N A A A N t t t 1 1 ......       4.3 A t = Actual in t period N = count data are needed 3. Until here, calculate about forecasting was finish, and the next step, analysis to make sure forecasting used Mean Absolute Deviation MAD 4. so, the Calculate the error actual data – forecasting data 5. Change the error by absolute data, the absolute error is move symbol front of numbers 6. Cumulative is calculation absolute error 7. The mad calculate is number cumulative in the last number divided count of all period The last step is calculated the mad and low value it is will be better model to forecasting. Finally the best model. Table 4.22 Comparison MAD Calculation MAD ARIMA MA=4 MAD ARIMA MA=5 MAD Exponential Smoothing Conclution Production 45,068,789 43,439,905 35,956,177 Exponential Smoothing Generated 2,928,446 2,822,606 2,336,334 Exponential Smoothing Consumption 2,796,825 2,672,064 2,248,789 Exponential Smoothing Population 619,761 613,684 157,288 Exponential Smoothing Calculate the error to make a sure of result if compare with result of forecasting of SPSS or EViews untill 2022 year by year from 2011-2022 by this formultion: 100 SPSS by g Forecastin EViews or Formula by g forecastin - SPSS by g Forecastin   Q 4.5 Example: error ratio between manual Excel and SPSS perpustakaan.uns.ac.id commit to user 40 Table 4.23 Error ratio between manual Excel and SPSS Years Production Generated Consumption Population 2011 1,89 1,89 0,02 0,31 2012 3,25 3,25 0,02 0,59 2013 4,48 4,48 0,02 0,85 2014 5,61 5,61 0,02 1,09 2015 6,64 6,64 0,02 1,31 2016 7,59 7,59 0,02 1,51 2017 8,46 8,46 0,02 1,70 2018 9,27 9,27 0,03 1,87 2019 10,03 10,03 0,03 2,03 2020 10,73 10,73 0,03 2,17 2021 11,38 11,38 0,03 2,30 2022 11,99 11,99 0,03 2,41 commit to user 41

CHAPTER V CONCLUSION AND RECOMENDATION

5.1 Conclusion

Based on the research and the result of analysis about forecasting of the electricity demands in Libya using time series Stochastic method for long-term from 2011-2022, we can conclude the following: We were able to determine the long-term demands for electricity in Libya in term of data from 2000-2010, by using forecasting time series. Time series forecasting method is formulated with time series data. Technical time series: there are two time series, they are deterministic and stochastic time series. Time series deterministic is a forecasting method with time series data where the data must meet the data stationary. This is due to a deterministic time series forecasting are methods appropriately. Terms of stationary data are used as the basis for forecasting the proper method deterministic time series. Studying energy demands at a disaggregated sectoral level for policy purposes. An application to the demand for electricity in Libya shows that alternative policy options for market reforms can be based on reliable long-term forecasts from in-sample parameter estimates of long-run relationships that are useful for decision-making To provide mathematical data that can be used as consideration in deciding a particular policy in the field of electricity supply, mathematical data that can be used in A stochastic time series is forecasting method assuming there is still a possibility of committing a mistake. The data used for the analysis of stochastic time series does not require stationary requirements. The research analysis process is a stochastic time series analysis because the data used does not meet the requirements of stationary. On the results of stochastic time series analysis using SPSS, the process runs smoothly. Models found in the SPSS analysis is ARIMA. The results of the analysis of SPSS can be presented. As for the mathematical models that can be used as a material consideration in deciding a particular policy in the field of power supply, perpustakaan.uns.ac.id commit to user