The Effect of Phase on Paper Thickness Optimization in the Recycling Process Using Recycling Machine

Efforts to manage the recycling of paper waste into new paper have been carried out in recent times. It takes a tool or machine that is able to effectively and efficiently recycle used paper into new paper. There are several factors that affect the effectiveness of paper recycling machines, one of which is the paper thickness. One method that can be used to analyze the factors that influence paper thickness in the paper production process using a paper recycling machine is regression analysis. Regression analysis is data analysis techniques in statistics that is used to examine the relationship between several independent variables and dependent variable. However, if we want to examine the relationship or effect of two or more independent variables on a dependent variable, the regression model used is a multiple linear regression model. This study purposes are to analyze the factors that influence paper thickness using a paper recycling machine using multiple linear regression and to inform the modeling about that. The results showed that the factors that affect the paper thickness optimization are destruction and press phase. Kata Kunci:: Multiple linear regression analysis; Paper making phase; Paper recycling machine; Paper thickness Abstract Upaya pengelolaan daur ulang sampah kertas menjadi kertas baru telah banyak dilakukan pada jaman sekarang. Dibutuhkan suatu alat atau mesin yang mampu secara efektif dan efisien dalam mendaur ulang kertas bekas menjadi kertas baru. Terdapat beberapa faktor yang mempengaruhi tingkat efektifitas mesin daur ulang kertas diantaranya adalah ketebalan kertas. Salah satu metode yang dapat digunakan untuk menganalisis faktor-faktor yang mempengaruhi ketebalan kertas pada proses produksi kertas menggunakan mesin daur ulang kertas adalah analisis regresi. Analisis regresi merupakan teknik analisis data dalam statistika yang digunakan untuk mengkaji hubungan antara beberapa variabel bebas dengan variabel tidak bebas. Namun, jika ingin mengkaji hubungan atau pengaruh dua atau lebih variabel bebas terhadap satu variabel tidak bebas, maka model regresi yang digunakan adalah model regresi linier berganda. Tujuan dalam penelitian ini yaitu menganalisis faktor-faktor yang mempengaruhi ketebalan kertas menggunakan mesin daur ulang kertas menggunakan regresi linier berganda serta memberikan informasi pemodelan mengenai hal tersebut. Hasil penelitian menunjukkan bahwa faktor yang mempengaruhi keoptimalan ketebalan kertas adalah fase penghancuran dan pemadatan kertasUpaya pengelolaan daur ulang sampah kertas menjadi kertas baru telah banyak dilakukan pada jaman sekarang. Dibutuhkan suatu alat atau mesin yang mampu secara efektif dan efisien dalam mendaur ulang kertas bekas menjadi kertas baru. Terdapat beberapa faktor yang mempengaruhi tingkat efektifitas mesin daur ulang kertas diantaranya adalah ketebalan kertas. Salah satu metode yang dapat digunakan untuk menganalisis faktor-faktor yang mempengaruhi ketebalan kertas pada proses produksi kertas menggunakan mesin daur ulang kertas adalah analisis regresi. Analisis regresi merupakan teknik analisis data dalam statistika yang digunakan untuk mengkaji hubungan antara beberapa variabel bebas dengan variabel tidak bebas. Namun, jika ingin mengkaji hubungan atau pengaruh dua atau lebih variabel bebas terhadap satu variabel tidak bebas, maka model regresi yang digunakan adalah model regresi linier berganda. Tujuan dalam penelitian ini yaitu menganalisis faktor-faktor yang mempengaruhi ketebalan kertas menggunakan mesin daur ulang kertas menggunakan regresi linier berganda serta memberikan informasi pemodelan mengenai hal tersebut. Hasil penelitian menunjukkan bahwa faktor yang mempengaruhi keoptimalan ketebalan kertas adalah fase penghancuran dan pemadatan kertas


INTRODUCTION
Manufacturing companies are companies engaged in processing raw materials by spending other costs into finished goods that are ready for sale. Manufacturing companies throughout the world are affected by globalization and are forced to become more competitive every day to maintain their profitability (Goddard, Tavakoli, & Wilson, 2005;Rahmasari, 2017). Some manufacturing companies are engaged in processing waste that no longer has the benefit of being a recycled product so that it can be reused and useful for life.
Lack of awareness from humans results in a lot of littering. It can cause some pollution. However, all can be overcome if we are creative in utilizing the garbage. Besides being able to preserve the environment, it can also increase our income. One of the waste utilization techniques is recycling by making various kinds of crafts and new products (Kementerian Keuangan Republik Indonesia, 2019).
The type of waste that is commonly found is paper waste. Paper waste as wasted goods can be used by recycling into unique and artistic paper or even becoming new paper for reuse. Through paper recycling, in addition to dealing with environmental problems, can also be used as an alternative in creating new job fields (Asmatulu & Asmatulu, 2011;Yafie, Kustiawan, & Seken, 2018;Zahra & Damanhuri, 2011). Efforts to manage the recycling of paper waste into new paper have been carried out in recent times. It takes a tool or machine that is able to effectively and efficiently recycle used paper into new paper. There are several factors that affect the effectiveness of paper recycling machines, one of which is the paper thickness. Therefore, it is necessary to do research on what factors influence the thickness of the paper in order to create optimal results in the production of paper on a paper recycling machine (Faraz, Biermann, & Weinert, 2009). One method that can be used to analyze the factors that influence paper thickness in the paper production process using a paper recycling machine is regression analysis.
Regression analysis is data analysis techniques in statistics that is used to examine the relationship between several independent variables and dependent variable (Effendi, Mursilah, & Mujiono, 2018;Widiyawati & Setiawan, 2015). If we want to examine the relationship or effect of one independent variable, the regression model used is a simple linear regression model. However, if we want to examine the relationship or effect of two or more independent variables on a dependent variable, the regression model used is a multiple linear regression model. Then to get a simple linear regression model and multiple linear regression models can be obtained by estimating its parameters using certain methods. The methods that can be used to estimate the parameters of a simple linear regression model or multiple linear regression model is ordinary least square (OLS) and maximum likelihood estimation (MLE) (Kutner, Nachtsheim, Neter, & Li, 2005).
This study purposes are to analyze the factors that influence paper thickness using a paper recycling machine using multiple linear regression and to inform the modeling about it to get the optimal paper thickness.

Multiple Linear Regression
The general form of multiple linear regression model with -independent variable as in the following equation (Johnson & Bhattacharyya, 2010).

Figure 1. Results of data exploration
The normal QQ plot shows a normal spread error, this is seen from the points that are around the line. Error plot vs fitted value shows a homogeneous errors range, it can be seen from the absence of patterns formed and the same bandwidth. Error plot vs order shown no pattern, so it can be concluded that the errors are mutually independent.

Figure 2. Result of Regression Analysis
Interpretation: The results of regression analysis for with 1 to 6 produced significant F test (P-value < 0.05), so there is sufficient evidence to state that there is at least one explanatory variable affecting . The t-test on each explanatory variable concluded that the 1 and 6 had P-value < 0.05, it means that there is sufficient evidence to state that both variables affect at the real level 5%, while the other explanatory variables do not affecting . The result of 2 value is 73.3%, it means that the variance of that can be explained by 1 to 6 is 73.3%, while the remaining 26.7% is explained by variables outside the model. This 2 value is relatively small, so it can be concluded if the model formed is not good enough, this is also supported by the 2 value = 54.7% which is relatively low. From this regression analysis, no outliers or observations were found to be influential. Glejser test produced P-value = 0.193, so there is sufficient evidence to state that the error variance is heterogenic at the real level 5%, in other words, the assumption of homogeneous error variance is fulfilled. Interpretation: Hypothesis: 0 : independent mutual error 1 : dependent mutual error Conclusion: Runs Test produced P-value = 0.129, is means there is not sufficient evidence yet to state that the error not mutually independent at the real level 5%, so the assumption of error independencies is fulfilled. The assumption of multicollinearity can be seen through the value on each independent variable. It can be seen that there is no value greater than 9, so it can be concluded that there is no multicollinearity. So this assumption is fulfilled.

Result of detect of multicollinearity
Regression analysis with 1 to 6 produce 2 value is 73.3% and there is no violation of assumption. Although the model formed is good enough, but not all variables are significant, so it is necessary to create a new model that can explain to be better. Some ways are used best subset and stepwise regression. The results of best subset and stepwise regression are as follows:

Figure 7. Result of best subset
Interpretation: The result of best subset concluded that the variables 1 and 3 were included into the model. When 1 and 3 are included, the result of 2 value model is quiet large compared to the other models, which is 70.8%. Cp Mallow value of 1.1 can also be said to be good because it approaches the number of explanatory variables included into the model that is 2. The value of 6.8268 is relatively low when compared by the other models. Stepwise regression suggested that the variables 1 and 3 were included into the model.
Base on the results of best subset and stepwise regression, it can be decided to use variables 1 and 3 into the model. And the result of regression analysis with 1 and 3 as follows: Figure 9. Result of data exploration after best subset and stepwise regression Interpretation: The normal QQ plot shows a normal spread error. Error plot vs fitted value shows a homogeneous errors range, it can be seen from the absence of patterns formed and the same bandwidth. Error plot vs order shown no pattern, so it can be concluded that the errors are mutually independent. Figure 10. Result of regression analysis after best subset and stepwise regression Interpretation: The result of regression analysis produces the formula: = 9.87 + 0.644 1 + 0.211 2 (6) F test produced the testing statistics of 32.74 and P-value = 0.000, so it can be concluded there is sufficient evidence to state that there is at least one explanatory variable that affect at the real level 5%. Following the F test, the test of each explanatory variable using t test produced P-value = 0.000 and 0.128 for variables 1 and 3 , so it can be concluded there is sufficient evidence to state that 1 affects while 3 does not affect at the real level 5%. The result of 2 value is 70.8%, it means that the variance of that can be explained by model is 70.8%, while the remaining 29.2% is explained by variables outside the model. 2 and 2 values on this model are larger than previous model, in other word this model is better. Regression analysis also shows the observed effect on the data, namely the 24th observation. Because this observation is influential, this observation is not removed from the data. Although 3 is not significant, but this model is better than when is only regressed by 1 (it can be seen from best subset).  Interpretation: value on the variables 1 and 3 are 1.553 and 1.553 indicate that there is no correlation between explanatory variables, so the assumption of the absence of multicollinearity is fulfilled.

Analysis Result
Based on the results of multiple linear regression analysis and after proven by the error assumption test, obtained the most optimal model in producing paper thickness in the recycling process using paper recycling machine is = 9.87 + 0.644 1 + 0. 211 3 From the model above, it can be seen that the phases that most influence the paper thickness are first phase ( 1 ) namely destruction phase and ( 3 ) namely press phase.

CONCLUSION
The phases that influence the paper thickness on the recycling process using paper recycling machine are first phase ( ) namely destruction phase and third phase ( ) namely press phase. This is reinforced by the data obtained that has fulfilled the error assumption test. So that in producing optimal recycled paper thickness, the phase that needs attention are destruction phase and press phase.