PREDICTION FOR TOTAL NUMBER OF LAB PARTICIPANTS BY FUZZY TIME SERIES METHOD ( CASE STUDY : INFORMATION ENGINEERING OF BHAYANGKARA SURABAYA UNIVERSITY )

Forecasting is a way to estimate a future value with using past data. One method of forecasting is the fuzzy method time series. The purpose of this study is to predict the number of students practitioners follow Department of Informatics University Bhayangkara Surabaya by using fuzzy method time series. The created app can be used to predict the next 1 year. If the actual data in the year predicted inputted, the application can predict the next year again. The prediction error rate is calculated using Mean Absolute Percentage Error (MAPE). From the test results in predicting the number of students followers 7 courses Practicum Informatics Engineering Bhayangkara University of Surabaya in 2010-2012 using the method proposed in this thesis for practicum PTI obtained MAPE value of 20.50%, Practical ANP obtained MAPE value of 0.50%, Network Computer practicum obtained MAPE value at 8.50%, practicum Database obtained MAPE value of 0.50%,Managemen Network Computer practicum obtained MAPE value of 14.50%, practicum PKG obtained MAPE value of 0.84% and practicum PBO obtained MAPE value of 0.21%. Based on the results of testing the data it can be concluded that the fuzzy time series method when used on more data many, it will get the accuracy of better and precise forecasting values.


INTRODUCTION 1.Background
In human life always faced with several options, where with the forecasting can help in predicting activity on life in the future.Forecasting is a process which aims to predict in between some alternative actions will concluded so that can be expected in helps to determine the outcome.Forecasting for decision making as well experienced by the Faculty of Informatics Engineering Bhayangkara University in determining participants of practicum followers.Lack of information on the number of students who follow the practicum and the student already complete the requirements of the lab, so difficult in determining many students who follow the lab.For avoid it, then it is needed a system that can help the part academic in solving deep problems predict and determine follower followers practice.The student forecasting process following the lab can be predicted with the availability of more than one option possibilities.Then used forecasting with the method Fuzzy Time Series that can help in decision making on the best and worst possible in order to obtain good results meet certain criteria.Forecasting with the Fuzzy Time Series method is based off data of applicants of the practicum obtained from Lecturer Chairman of Engineering Praktikum Department Informatics of Bhayangkara University with proof of payment from bank and KRS data college student.With that method expected results can help determine the participants who follow practicum Informatics Engineering Department.

Purpose
This research aims to apply the Fuzzy Time Series method to helping to predict the number of students participants of practicum Engineering followers Informatics of Bhayangkara University Surabaya, so it can be determined on a regular basis easy to see the level of accuracy forecasting.

METHODOLOGY 2.1. Literature review
There are several previous studies which uses the Fuzzy Time method Series, such as Averange-Based Fuzzy Time Series for exchange rate forecasting foreign exchange (USDIDR Exchange Rate Case Study and EUR-USD) [11], forecasting the percentage change of data Composite Stock Price Index (CSPI) With Fuzzy Time Series [10], forecasting the sales method of Fuzzy Ttime Series Ruey Chyn Tsaur [12], Application of Fuzzy Time Series By Elliott Wave Principle For Stock Price Prediction [14], Forecasting Number of Candidates Student Stmik Duta Bangsa Using Time Invariant Fuzzy Time Series Method [13], Forecasting System Movement of Capital Market Index With The Fuzzy Time Series Method.The level 0 chart will be decomposed into several sub-process level diagrams next called the level 1 diagram.There are several sub processes in the diagram next flow among others Logging, Manage Master Data, Calculation of Forecasting Method FTS and FTS Forecasting Testing.There is some data store of them is user, students, practicum, FTS forecasting, and FTS testing.The diagram in figure 1 shows process level diagram 1.

Figure 2. DFD Level 1
Any process found on the level 1 diagram will be decomposed again into sub-chapters of level diagram process next.Login Sub at level 2 will be decomposed into 3 processes and 1 data restore.3 process consists of insert, update user data and Access Rights checking.While the data restorenya is Data user.Figure 2 shows sub login process.DFD level 2 is a depiction from each process that exist on level 1 DFD, is DFD level 2 login process (Figure 3), and DFD level 2 manage master data (Figure 4).

Fuzzy Time Series Method
Steps in forecasting participants using practicum followers method of Fuzzy Time Series in this research are as follows (Hsu et al. 2010): 1.The set of the universe.
The universal set U = [ Dmin , Dmax ] determined according to historical data exist, and divide it into an odd number of sub-intervals with equal width intervals.2. Fuzzification Process A1 , A2 , ... , Ak is a fuzzy sets that are the linguistic variable is determined according to the state of the universe, at where k is the number of intervals obtained from the first step then define the set of sets The fuzzy according the following models (Song and Chissom): is a degree deep uk interval membership fuzzy set At k = 1, a fuzzy set is obtained (fuzzy set of student numbers the fewest).At k = n, obtained (fuzzy set of numbers student of the most).The greater the value of k, the fuzzy set the number of students will move from the least become the set fuzzy number of students the most many.

Second-order fuzzy logical relationship
If the result of fuzzification number of students in the year i-2 is, amount student in the year i-1 is then the number of students in year I is, where, as a side left relationship is referred to as current state and Ak as the right side relationship referred to as next state.Fuzzy logical relationship group formed by dividing fuzzy logical relationship that has been obtained into sections based on the left side of the fuzzy logical relationship (current state).

Defuzzification process
The defuzzification process changes a fuzzy quantities become strict quantities.The output in this process is a forecasting value determined using the following rules: (1) If in the group is obtained exactly one next state, as fuzzy logical relationship follows: where is the maximum degree of degree of membership from AK at intervals UK , and midpost (middle value) of UK is AK , then the forecasting value for the group in question is Mk.
(2) If in the group get more from one next state, as fuzzy logical relationship follows: AI , AJ → Ak1 , Ak2… Akn where the maximum value of degree membership from at Ak1 , Ak2… Akn intervals Uk1 , Uk2… Ukn and midpost (middle value) of Uk1 , Uk2… Ukn is mk1 , mk2… mkn then forecasting value for the group is (mk1,+ mk2… + Ukn)/n (3) If the group is not found next state, as fuzzy logical relationship follows: AI , AJ → # where # denotes unknown value and maximum value of degree of membership from AI and AJ available at intervals ui and uj and and midpost (middle values) of ui and ui is MI and MJ then the forecasting value for that group is MJ + ((MJ MJ) / 2).

Forecast rules
This stage consists of two parts, namely matching part (current state of fuzzy logical relationship group) and forecasted value.Determination of forecast value is determined by matching current state fuzzy logical relationship year i with matching part.If current state with rules which has formed a match, then forecast value in the same year as forecast value of the matching part concerned.

Data of practicum participants Jarkom
By whole, results process forecasting the data can be seen on table 3  Then tested the data for analyze accuracy method that used by generating a MAPE minimum.From Table 3.10 can be seen that in forecasting the number of students followers of Jarkom practice by method fuzzy time series got the total value of MAPE by 8.50%.
4. Data of participants of Database practicum By whole, results process such data forecasting can be seen in Table 3  Then tested the data for analyze the accuracy of that method used by generating MAPE the minimum.From Table 3.12 it can be seen that in forecasting amount student of Database practicum followers with the fuzzy time series method obtained a total value of MAPE of 0.50%   Then tested the data for analyze accuracy method that used by generating a MAPE minimum.From Table 3.16 it can be seen that in forecasting the number of student followers practicum PKG with fuzzy time method series obtained a total value of MAPE of 0.84%.

Data of PBO practicum participants
By whole, results process forecasting the data can be seen on table 3

Analysis of Results
Based on data forecasting results of the eye college practicum, it appears that if there is empty data (zero) and too few periods, then the amount of data entered into forecasting process with fuzzy time method series shows less results maximum.As for forecasting on the comparison data is the number of students the practicum in the Reproduction Laboratory Livestock Faculty of Animal Husbandry University Brawijaya can still show results better forecasting than data participants of Informatics Engineering Bhayangkara University Surabaya.On testing the measurement process data forecasting error rate followers practice majors Technique Informatics University Bhayangkara Surabaya where the result of error measurement forecasting is greater than results in data comparison of the Number of Student data lab in the Reproduction Laboratory Livestock Faculty of Animal Husbandry University Brawijaya year 2010-2014.

CONCLUSION
From the discussion that has been done then it can be determined a conclusion as follows : 1) From the process of calculating the forecasting done to the practicum majors

Table 3 .
1 Practical Fuzzification Data PTI  Second-order fuzzy logical relationship Based on fuzzification results on the second step, can be determined second order fuzzy logical relationship that can be seen on table 3.2 and table 3.3

Table 3 .
2 Second-order fuzzy logical relationship PTI Practicum

Table 3 .
3 Second-order fuzzy logical relationship group PTI Practicum  Defuzzification process Group 1 calculation, from Table 3.3 as follows : A21, A12 → A1 where the membership value is maximumvfor the fuzzy set A1 falls on interval u21 = [210, 220], and value the middle of the interval of u21 is 215 then forecasting value for group 1 is 215.
 Forecast rules Based on defuzzification results on the fourth step, can be determined some rules that can be seen on table3.4 and has been obtained in the table 3.5.

Table 3 .
6 Testing MAPE data PTI lab workThen tested the data for analyze accuracy method that used by generating a MAPE minimum.From Table3.6 it can be seen that in forecasting the number of student followers practicum PTI with fuzzy time method series obtained a total value of MAPE of 20.50% 2. Data of ANP practicum participants By whole, results process forecasting the data can be seen on table3.7

Table 3 .
8 MAPE Testing of practicum data ANP

Table 3 .
9 Application of fuzzy time method series Practicum Jarkom

Table 3 .
.11. 11 Application of fuzzy time method series Database Practicum

Table 3 .
12 MAPE Testing of practicum data Database

Table 3 .
5. Data of Manjarkom practicum participants By whole, results process forecasting the data can be seen on table3.13.13 Application of the fuzzy time series method Manjarkom Practicum

Table 3 .
14 MAPE Testing of practicum data ManjarkomThen tested the data for analyze accuracy method that used by generating a MAPE minimum.From

Table 3 .
14it can be seen that in forecasting the number of student followers Manjarkom practice with fuzzy method time series obtained the total value of MAPE of 14.50% 6.Data of PKG practicum participants By whole, results process forecasting the data can be seen on table3.15.

Table 3 .
15 Application of fuzzy time method series PKG Practicum

Table 3 .
17 Application of the fuzzy time series method PBO Practicum

Table 3 .
18 MAPE Testing of practicum data PBOThen tested the data to analyze the accuracy of that method used by generating a MAPE minimum.From Table3.18 it can be seen that in forecasting the number of students follower of PBO with fuzzy method time series obtained the total value of MAPE of 0.21%.Experiment as a comparison analysis performed on the Number of Student data lab in the Reproduction Laboratory Livestock Faculty of Animal Husbandry University Brawijaya year 2010-2014, with data as follows :

Table 3 .
19 Data Number of Students in practicum Reproduction Laboratory Results forecasting of 2 courses practicum data Number of Students who practicum at Livestock Reproduction Laboratory Faculty Universitas Brawijaya Farming in 2010 -2014, can be seen in table 3.20 and table 3.21.Then the test results can be seen in Table 3.46 and Table 3.47.

Table 3 .
20 Application of the fuzzy time series method Number of students of Biotechnology practicum

Table 3 .
21 Testing MAPE data Amount Student of Biotechnology practicum Year Number of students Fuzzy logical relationship Matched Rule No. Forecasting

Table 3 .
22 Application of the fuzzy time series method Number Amount Student of Livestock Reproduction Laboratory

Table 3 .
23 Testing MAPE data Amount Student of Biotechnology practicum Technique Informatics University Bhayangkara Surabaya from the year 2010 -2012 obtained forecasting results on year 2013 for practical subjects PTI amounted to -40, practical subjects ANP of -60, course practicum JARKOM for -20, course Database practicum of -25, eyes Manjarkom's practicum course is -20, PKG practicum course is 135 and PBO practicum courses are as big as -5.While on data comparison data Number of students who practicum at Laboratory Reproduction Livestock Faculty Farms University Brawijaya year 2010-2014 obtained forecasting