Prediction of Safety Stock Using Fuzzy Time Series ( FTS ) and Technology of Radio Frequency Identification ( RFID ) for Stock Control at Vendor Managed Inventory ( VMI )

This research was conducted by prediction of safety stock using Fuzzy Time Series (FTS) and technology of Radio Frequency Identification (RFID) for stock control at Vendor Managed Inventory (VMI). Well-controlled stock influenced company revenue and minimized cost. It discussed about information system of safety stock prediction developed through programming language of PHP. Input data consisted of demand got from automatic, online and real time acquisition using technology of RFID, then, sent to server and stored at online database. Furthermore, data of acquisition result was predicted by using algorithm of FTS applying universe of discourse defining and fuzzy sets determination. Fuzzy set result was continued to division process of universe of discourse in order to be to final step. Prediction result was displayed at information system dashboard developed. By using 60 data from demand data, prediction score was 450.331 and safety stock was 135.535. Prediction result was done by error deviation validation using Mean Square Percent Error of 15%. It proved that FTS was good enough in predicting demand and safety stock for stock control. For deeper analysis, researchers used data of demand and universe of discourse U varying at FTS to get various result based on test data used.


Introduction
Stock management is one of production process planning and controls whose purpose is to decrease total cost of stock material and stock level during lead time and acquisition cost.Management developing stock policy which can minimize operational total cost is the main purpose of planning and control.Stock management is an important factor in production process, one factor influencing stock management is demand prediction; demand fluctuation influences product stock and production activity greatly [1].
An important component of chained supply management is stock management.Stock management can spotlight prediction mistake and decision policy depending on demand having potential to prediction mistake.Prediction having the greatest influence to final user's decision can be used to develop demand prediction concept which can give significant influence to the improvement of a company profit [2].
Demand prediction in control management and production supply becomes interesting challenge to be researched because most of them work on data of time series as having been done to overcome problem prediction, like prediction in information system management, health care, economy prediction, selling prediction, budgeting analysis, stock exchange fluctuation, and business analysis, etc [3].
Fuzzy time series (FTS) can design problem of prediction having linguistic value with information having been long time.FTS also can use more observation in prediction having been applied to overcome non-linear.Based on theory of fuzzy compilation, FTS model came from Song and Chissom in 1993, FTS was used to predict the registration of Alabama University.Chen presents new model by using simple fuzzy relation and simple arithmetic calculation [4][5].
Fuzzy time series can predict product need for the next period and this prediction can be arranged based on time period needed.By integrating fuzzy time series to an information system to calculate ROP score of each product, the error average of ROP score got after being examined by using method of Average Forecasting Error Rate (AFER) was 7,13%.Fuzzy times series can predict the number of stock needed in stock room, report stock availability, and give goods stock information so high economy efficiency is got [6].Time series is an ordered time series arranged from quantitative individual characteristics or collective phenomenon taken from time period successively.To understand time series characteristic, many researchers E3S Web of Conferences 31, 11005 (2018) https://doi.org/10.1051/e3sconf/20183111005ICENIS 2017 have adopted, analyzed, and developed time series method whose final purpose is to find pattern or formula that can be used to predict the future [7].
Radio Frequency Identification (RFID) technology is one technology used in supply chain management using modern.By using wireless technology, a company can track RFID tags easily without physical contact.RFID technology has been proven to be very useful in planning of production, transportation, and warehousing [8] RFID can integrate into company business process so that it is possible for every entity marked can communicate with all organization information infrastructure, so it can enhance information of supply chain.In business technology process, RFID shows that it can operate in small and middle retail industry and can describe effect of RFID in business operation [9] Vendor Managed Inventory (VMI) has very significant benefit for supply chain and each company.VMI gives competitive profit to retailer related to higher product availability provided by suppliers with the chance to increase production and marketing efficiency.VMI can increase fulfillment frequency with a small number and decrease stock level for all involved in distribution and supply chain.VMI can optimize supply chain performance in which the producer is in charge to keep distributor's stock level.Producer has access to distributor's stock data and is in charge to order [10][11].

Data Acquisition
This acquisition data process on application of safety stock prediction using Fuzzy Time Series (FTS) and Radio Frequency Identification (RFID) technology for stock control at Vendor Managed Inventory (VMI) applies RFID censor technology which is censor detecting id tags put on the goods using radio waves and analyzed to be data time series stored at local database on microprocessor by using internet network, data of goods demand history is sent to web server and stored at online database then predicted by using fuzzy time series and used to determine safety stock.Data acquisition route is shown in Fig 1.

Modelling by using fuzzy time series
Predicting by using fuzzy time series model is method of data prediction using principles of fuzzy whose base is catching formula of long time data then used to project the future data.Modelling of prediction by using fuzzy time series has some steps as follows : ( (2) (5) 3. Dividing universe of discourse U with some series of data u1, u2, …. un and determining linguistic score.4. Doing fuzzification and fuzzy set from data of actual histories.5. Calculating score of fuzzy data of actual history by using the following pattern : 6. Choosing basis of model W which is very appropriate and calculating fuzzy using the following pattern: 7. Doing defuzzyfication of calculation result from the above step then, calculating prediction result by using the following pattern :

Results and Discussion
Data acquisition process on application of safety stock prediction using Fuzzy Time Series (FTS) and Radio Frequency Identification (RFID) technology for stock control at Vendor Managed Inventory (VMI) has some steps as follows: a. Acquisition process of goods demand data History data used to predict by using fuzzy time series is data of every month-actual demand data from PT. Quindo food, period of 2012 -2016, with 60 data as sample taken from scan tag id on goods using censor technology of RFID as in table 1.

Evaluation and validation of calculation result
From evaluation and validation of error deviation toward fuzzy time series above, error deviation also has been tested by using variance of number of universe of discourse starting from 3, 4, 5, 6 and 7 as well as data number starting from 12, 24, 36, 48 and 60 data so that the result got is shown in Figure 3.

Conclusion
Prediction by applying algorithm of Fuzzy time series done with variance of interval score toward universe of discourse and variance of data number can be applied to predict safety stock.It can be proven by testing result using data number of 60 and the average error score got was 15% measured by using method of Mean Absolute Percentage Error (MAPE).Prediction result accuracy is influenced by data number fluctuation, the size (small and big) of interval score of universe of discourse and minimal and maximal score of universe of discourse.

Fig. 1 .
Fig. 1.Acquisition process route of goods demand data

1 .
Defining universe of discourse U until fuzzy set can be determined as U = [x, y]. 2. Determining minimal and maximal score of actual history data (Xmin = x, Xmax = y).

Table 1 .
Data of actual demand

Table 2 .
Data of prediction result