Please cite this article in press as: Atalay KD, et al. A hybrid algorithm based on fuzzy linear regression analysis by quadratic programming for time estimation: An experimental study in manufacturing industry. J Manuf Syst (2014), http://dx.doi.org/10.1016/j.jmsy.2014.06.005 ARTICLE IN PRESS G Model JMSY-303; No. of Pages 7 Journal of Manufacturing Systems xxx (2014) xxx–xxx Contents lists available at ScienceDirect Journal of Manufacturing Systems j ourna l h omepage: www.elsevier.com/locate/jmansys A hybrid algorithm based on fuzzy linear regression analysis by quadratic programming for time estimation: An experimental study in manufacturing industry Kumru Didem Atalay a,∗ , Ergün Eraslan b , M. Oya C ¸ inar c a Department of Medical Education, Baskent University, Ankara, Turkey b Department of Industrial Engineering, Yildirim Beyazit University, Ankara, Turkey c Department of Biomedical Engineering, Baskent University, Ankara, Turkey a r t i c l e i n f o Article history: Received 13 May 2010 Received in revised form 6 May 2014 Accepted 8 June 2014 Available online xxx Keywords: Fuzzy linear regression Quadratic programming Time study Standard time Time estimation Forward selection a b s t r a c t In time studies, estimation of the standard times with direct or indirect measurement methods is partic- ularly difficult in companies having complex production schedules or ones employing an inexperienced workforce. Such companies require new and specific time measurement procedures. In this study, a new time estimation algorithm based on fuzzy linear regression analysis (FLRA) by quadratic programming (QP) is proposed for specific manufacturing systems. In our study, data is provided by one of the biggest casting and machining companies in Europe. The database includes items that have similar production processes. A fuzzy linear regression model is built by using the previously measured standard times of a product family. The model developed is used for estimating the standard times of the remaining products. FLRA based on QP approach facilitates integration of the central tendency of least squares and possible properties of fuzzy regression. The main factors that directly impact standard times are determined and used for the estimation of the fuzzy standard times. Through utilization of sum of squares error (SSE) and index of confidence (IC), the important factors in the model are identified. The use of QP makes it pos- sible to reconcile the minimization of the deviation of central tendency and the spreads of membership functions in a simultaneous manner. In this study, the efficiency of the proposed algorithms in casting companies is authenticated. Besides, it is seen that this estimation procedure could be implemented easily for various sectors using the relevant algorithms. © 2014 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved. 1. Introduction In recent years, due to the exacerbating competition in the market, determination of the exact standard time of manufactur- ing products has become an essentiality [1]. It is very difficult to prepare manufacturing schedules, short and long term forecasts, capacity planning, pricing and some other technical and managerial activities in a company without accurate standard time. Unfor- tunately, direct or indirect work measurement procedures, e.g., time study, activity sampling, standard data synthesis, analytical estimation, comparison, prediction and elementary motion stan- dards are insufficient in most cases to determine the exact standard ∗ Corresponding author at: Department of Medical Education, Baskent University, Tas ¸ kent cad. 77, Sok. No: 11, 06490 Bahc ¸ elievler, Ankara, Turkey. Tel.: +90 312 212 90 65x2238; fax: +90 312 221 37 59. E-mail addresses: katalay@baskent.edu.tr (K.D. Atalay), eraslan@ybu.edu.tr (E. Eraslan), oyacinar@baskent.edu.tr (M.O. C ¸ inar). time. For example, the cost of time study, as well as inadequate environmental conditions, can be prohibitive due to complex pro- duction processes. Thus, new and effective methods are needed to be developed regarding this issue. Some of the recent studies in the literature are outlined below. Koelling and Ramsey [2] studied the effects of multimedia in developing and applying work measurement methods, Cohen et al. [3] examined successful integration of automatic speech recogni- tion (ASR) into industrial systems and found that the generation of time standards is a time consuming process that always slows down the work measurement task and increases cost. By using ASR, 70% time reduction is achieved. Focusing on engineers using 100 indus- trial engineering programs in US, Freivalds et al. [1] tested the effect of work measurement and design systems on customer satisfac- tion. Praszkiewicz [4] presented the application of ANNs in small scale machining production, with a purpose similar to the study of Freivalds. Besides Eraslan [5] studied on the ANN method for standard time estimation of a specific manufacturing facility. Apart from the fact that there are some studies on the literature which use http://dx.doi.org/10.1016/j.jmsy.2014.06.005 0278-6125/© 2014 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.