A churn prediction model for prepaid customers in telecom using fuzzy classifiers

The incredible growth of telecom data and fierce competition among telecommunication operators for customer retention demand continues improvements, both strategically and analytically, in the current customer relationship management (CRM) systems. One of the key objectives of a typical CRM system is to classify and predict a group of potential churners form a large set of customers to devise profitable and targeted retention campaigns for keeping a long-term relationship with valued customers. For achieving the aforementioned objective, several churn prediction models have been proposed in the past for the accurate identification of the customers who are prone to churn. However, these previously proposed models suffer from a number of limitations which place strong barriers towards the direct applicability of such models for accurate prediction. Firstly, the feature selection methods adopted in majority of the past work neglected the information rich variables present in call details record for model development. Secondly, selection of important features was done through statistical methods only. Although statistical methods have been applied successfully in diverse domains, however, these methods alone without the augmentation of domain knowledge have the tendency to yield erroneous results. Thirdly, the previous models have been validated mainly with benchmark datasets which do not provide a true representation of real world telecom data consisting of noise and large number of missing values. Fourthly, the evaluation measures used in the past neglected the True Positive (TP) rate, which actually highlights the ability of a model to correctly classify the percentage of churners as compared to non-churners. Finally, the classifiers used in the previous models completely neglected the use of fuzzy classification methods which perform reasonably well for data sets with noise. In this paper, a fuzzy based churn prediction model has been proposed and validated using a real data from a telecom company in South Asia. A number of predominant classifiers namely, Neural Network, Linear regression, C4.5, SVM, AdaBoost, Gradient Boosting and Random Forest have been compared with fuzzy classifiers to highlight the superiority of fuzzy classifiers in predicting the accurate set of churners.     

Highly Tunable Thiol-Ene Photoresins for Volumetric Additive Manufacturing

Volumetric additive manufacturing (VAM) forms complete 3D objects in a single photocuring operation without layering defects, enabling 3D printed polymer parts with mechanical properties similar to their bulk material counterparts. This study presents the first report of VAM-printed thiol-ene resins. With well-ordered molecular networks, thiol-ene chemistry accesses polymer materials with a wide range of mechanical properties, moving VAM beyond the limitations of commonly used acrylate formulations. Since free-radical thiol-ene polymerization is not inhibited by oxygen, the nonlinear threshold response required in VAM is introduced by incorporating 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO) as a radical scavenger. Tuning of the reaction kinetics is accomplished by balancing inhibitor and initiator content. Coupling this with quantitative measurements of the absorbed volumetric optical dose allows control of polymer conversion and gelation during printing. Importantly, this work thereby establishes the first comprehensive framework for spatial–temporal control over volumetric energy distribution, demonstrating structures 3D printed in thiol-ene resin by means of tomographic volumetric VAM. Mechanical characterization of this thiol-ene system, with varied ratios of isocyanurate and triethylene glycol monomers, reveals highly tunable mechanical response far more versatile than identical acrylate-based resins. This broadens the range of materials and properties available for VAM, taking another step toward high-performance printed polymers.     

Relationship of senescence of corn (Zea mays L.) stalk pith, cob parenchyma and first developed leaf tissues to RNA content and synthesis

In stalk parenchyma tissue of WF9 X 38-11 single corn (Zea mays L.), there was a per cell increase in RNA synthesis and a slight increase in total RNA as cells became older. In cob parenchyma tissue of WF9 X 38-11 single cross corn, both total RNA content and RNA synthesis per cell decreased with the age of cells. In first developed leaf tissue of FR43 X FR14A single cross corn, RNA synthesis increased steadily but only slightly over its life span. The pattern of RNA synthesis and destruction in senescing leaf tissue of seedlings and two sources of parenchyma tissue of maturing plants appeared to differ.     

Analysis of "total toxaphene" and selected single congeners in biota by ion trap HRGC-EI-MS/MS using congener-optimized parent ion dissociations

A method for the quantification of selected toxaphene congeners as well as "total toxaphene" was developed based on electron ionization (EI) tandem (MS/MS) ion trap mass spectrometry (MS) and a combination of fragment ion dissociations. Congeners were separated by high-resolution gas chromatography. Compared to conventional EI low-resolution MS, a 5-20-fold gain in sensitivity could be obtained for octa- or nonachlorinated compounds such as toxaphene #26 and #62 (according to Parlar nomenclature), allowing for their detection in the low picogram range in biota. In addition, response factors for important congeners such as #26, #32, #40/41, #42, #44, #50, and #62 deviated not more than a factor of 2, which is much less as compared to negative ion chemical ionization. This reduces the risk for systematic errors when determining total toxaphene on the basis of a limited number of reference compounds or the technical mixture. Furthermore, chlordanes and polychlorinated biphenyls did not interfere when applying the proposed MS/MS technique. The applicability of the method was tested by determining both total toxaphene and levels of selected congeners in six Arctic wildlife samples collected from Nunavut, Canada, as well as by repetitive analyses of the SRM 1588 certified reference material.     

Rapid Identification and Comparison of Compounds with Antioxidant Activity in Coreopsis tinctoria Herbal Tea by High‐Performance Thin‐Layer Chromatography Coupled with DPPH Bioautography and Densitometry

A simple and efficient method based on high‐performance thin‐layer chromatography coupled with 2,2‐diphenyl‐1‐picrylhydrazyl (DPPH) bioautography (HPTLC‐DPPH) was established for the screening and comparison of antioxidants in different parts of Coreopsis tinctoria herbal tea from different origins and other related herbal tea materials, which used Chrysanthemum morifolium cv. “Gongju” and “Hangju” in this study. Scanning densitometry after DPPH derivatization was applied for the determination of antioxidant capacities of isolated compounds in each sample. It is considered that ethanol extracts of C. tinctoria had stronger antioxidant activity and more characteristic bands than those of 2 compared samples, C. morifolium cv. “Gongju” and “Hangju.” Chemometric analysis results showed that the combination of hierarchical clustering analysis and principal component analysis based on determined antioxidant capacities could be used for the discrimination of different parts of C. tinctoria and C. morifolium. Results showed that 7 compounds made up the major contributions of antioxidant activity in C. tinctoria, including okanin, isookanin, marein, flavanomarein, 5,7,3′,5′‐tetrahydroxyflavanone‐7‐O‐glucoside, 3,5‐dicaffeoylquinic acid, and chlorogenic acid. Therefore, 7 compounds were identified as major antioxidant biomarkers for quality control of C. tinctoria. Results demonstrated that the established method could be applied for the identification of C. tinctoria, and were beneficial for the bioactivity‐based quality control of C. tinctoria.     

Deformation of drops in extensional viscoelastic flow

The deformation of nylon drops in polyethylene, with and without an interfacial agent, in an extensional flow has been studied. The presence of an interfacial agent reduces the size of the dispersed phase, and the deformation of the drop is reduced. An analysis is given, which accurately predicts the deformation for all values of the capillary number considered. The predicted and observed shapes are, however, only in agreement at low values of capillary number. Possible causes for this discrepancy are discussed. © 1996 John Wiley & Sons, Inc.     

Crack Initiation in Perforated Propellants under high strain rate impact conditions

The fracture behaviour of gun propellant grains during impact has been measured on a modified Hopkinson Bar which uses a projectile as the input bar. Evidence for the detection of initial minor crack formation in seven-perforated cylindrical specimens of single base, double base and triple base propellant is presented. Under compressive load (pressurization rates of 100 GPa/s-200 GPa/s), minor crack formation does not immediately result in large scale crack propagation or fragmentation of the specimens. The single base and triple base propellant are equally more resistant to crack initiation than the double base propellant when tested in either the end-on or side-on modes. Implications for munition survivability are discussed in terms of mechanical frictional effects caused during compressive loading of cracked propellant or compressive heating in propellant cracks during the ballistic cycle.     

Investigation of post-spinning stretching process on morphological, structural, and mechanical properties of electrospun polyacrylonitrile copolymer nanofibers

Electrospun polyacrylonitrile (PAN) copolymer nanofibers with diameters of ∼0.3 μm were prepared as highly aligned bundles. The as-electrospun nanofiber bundles were then stretched in steam at ∼100 °C into 2, 3, and 4 times of the original lengths. Subsequently, characterizations and evaluations were carried out to understand morphological, structural, and mechanical properties using SEM, 2D WAXD, polarized FT−IR, DSC, and mechanical tester; and the results were compared to those of conventional PAN copolymer microfibers. The study revealed that: (1) the macromolecules in as-electrospun nanofibers were loosely oriented along fiber axes; although such an orientation was not high, a small extent of stretching could effectively improve the orientation and increase the crystallinity; (2) most of macromolecules in the crystalline phase of as-electrospun and stretched nanofibers possessed the zig-zag conformation instead of the helical conformation; and (3) the post-spinning stretching process could substantially improve mechanical properties of the nanofiber bundles. To the best of our knowledge, this study represented the first successful attempt to stretch electrospun nanofibers; and we envisioned that the highly aligned and stretched electrospun PAN copolymer nanofibers could be an innovative type of precursor for the development of continuous nano-scale carbon fibers with superior mechanical strength.