Towards a new approach to predict business performance using machine learning

Financial ratio plays a crucial role in business performance prediction, but the ability of the decision maker to use this method in adjusting management strategy has been extensively ignored. In this paper we attempt to build a fuzzy chance constrained least squares twin support vector machine (FCC-LSTSVM) to predict the business performance through the financial ratios. Specifically, machine learning techniques are utilized to build the models and 796 listed companies in China are selected as the data set. We find that different efficiencies are performed for different models with the same industry and different effectiveness are shown for different predicting time periods with the same method. In addition, the predicting achievements of business performance depend on the types of industries. This paper has extent significance both in theoretical development and managerial practices.     

Chunking during serial learning by a pigeon: II. Integrity of a chunk on a new list

Can a group of items that a pigeon chunks on one list function as such on a second list? In Experiment 1, the ordinal position of the chunk was held constant across both lists. Following training on a list of colors and achromatic geometric forms (A----B----C----D'----E'), the integrity of the color chunk [A----B----C] was maintained on a list of five colors (A----B----C----F----G) even though the basis for establishing that chunk was eliminated. The integrity of the chunk [A----B----C] was also maintained on a new list of colors and forms (A----B----C----D*----E*). In Experiment 2, the ordinal position of a chunk established on list1 (A----B----C'----D') was changed on list 2. As shown by positive transfer between lists 1 and 2, the integrity of the chunks [A----B] and [C'----D'] was maintained on lists X'----A----B----Y' and X'----C'----D'----Y', respectively. Conversely, the heterogeneous list X'----B----C'----Y' took longer to learn than the original list.     

Positional cues in serial learning: The spin-list technique

To test the hypothesis that serial learning depends largely on the encoding and retrieval of position-to-item associations, we examined whether people can learn spin lists on which starting position is randomly varied across successive learning trials. By turning positional information from a reliable cue into a source of intertrial interference, we expected learning to be greatly impaired. Contrary to this hypothesis, we found that participants were only slightly worse at serial learning under spin conditions and that this impairment reflects a substantial increase in initiation errors coupled with a small increase in intertrial forgetting. These data show that participants can effectively use nonpositional cues when positional cues are unreliable.     

The simultaneous acquisition effect: simultaneous task learning inhibits memory for order

Delayed recall of a list of words learned simultaneously with other lists is superior to that of a list of words learned singly. A Brown-Peterson-like task was used to investigate this simultaneous acquisition effect from the perspective of the item-order distinction. It was hypothesized that simultaneous task learning would impede the encoding of order information but promote the encoding of item-specific information relative to single-task learning. The results of the first four experiments strongly supported the hypothesis that simultaneous task learning decreases the encoding of order information but provided no evidence that it facilitated the encoding of item-specific information. Additionally, Experiment 5 showed that the simultaneous acquisition effect did not occur in a mixed-list design, demonstrating an important boundary condition of the effect.     

Psychology of learning: a new approach to study behavior of Rhodnius prolixus stal under laboratory conditions

Conditioning methodologies associated with the psychology of learning are suggested as a new strategy to investigate behavior of the assassin bug Rhodnius prolixus, which is the main vector of Chagas disease in Venezuela. Chagas disease is the fourth leading cause of death in Latin America, as it causes severe chronic illness and approximately 43,000 deaths per year. To illustrate this strategy, two preliminary experiments are reported. In the first, Pavlovian conditioning was examined by pairing an olfactory conditioned stimulus with a temperature unconditioned stimulus. A temperature of 42 degrees C elicits a complex behavioral sequence in R. prolixus consisting of proboscis extension and crawling. Over the course of 12 training trials, this behavioral sequence was not elicited by an olfactory conditioned stimulus. In the second experiment, a latent inhibition paradigm was used to pre-expose R. prolixus to an olfactory conditioned stimulus before pairing the odor with temperature. Over the course of training, an effect of pre-exposure was found. Suggestions for research are discussed and potential conditioned and unconditioned stimuli identified.     

A new probe technique for the identification of serial learning processes

We describe a new serial learning probe technique which permits identification of functional stimuli for each individual item immediately following the trial on which it has been learned. Some empirical evidence indicating the validity of the serial probe technique is also presented. The potential superiority of the probe technique over alternative experimental methods for determining the basic processes involved in serial learning is noted.     

Implicit learning of what comes when and where within a sequence: The time-course of acquiring serial position-item and item-item associations to represent serial order

Much research has been conducted aimed at the representations and mechanisms that enable learning of sequential structures. A central debate concerns the question whether item-item associations (i.e., in the sequence A-B-C-D, B comes after A) or associations of item and serial list position (i.e., B is the second item in the list) are used to represent serial order. Previously, we showed that in a variant of the implicit serial reaction time task, the sequence representation contains associations between serial position and item information (Schuck, Gaschler, Keisler, & Frensch, 2011). Here, we applied models and research methods from working memory research to implicit serial learning to replicate and extend our findings. The experiment involved three sessions of sequence learning. Results support the view that participants acquire knowledge about order structure (item-item associations) and about ordinal structure (serial position-item associations). Analyses suggest that only the simultaneous use of the two types of knowledge acquisition can explain learning-related performance increases. Additionally, our results indicate that serial list position information plays a role very early in learning and that inter-item associations increasingly control behavior in later stages.     

Constructions: a new theoretical approach to language

A new theoretical approach to language has emerged in the past 10-15 years that allows linguistic observations about form-meaning pairings, known as 'constructions', to be stated directly. Constructionist approaches aim to account for the full range of facts about language, without assuming that a particular subset of the data is part of a privileged 'core'. Researchers in this field argue that unusual constructions shed light on more general issues, and can illuminate what is required for a complete account of language.     

The effect of simultaneous versus serial responding on children's ability to coordinate two dimensions

Previous research by Lidster and Bremner (1999) on young children's ability to coordinate two dimensions has shown that performance on construction tasks (in which children have to give the correct coordinates for a point in space that is already known) is superior to performance on interpretation tasks (in which children are given a pair of coordinates and have to locate the correct point in space on the basis of these). The present study investigates the suggestion made by Lidster and Bremner that construction tasks may be easier due to the possibility that they can be solved by attending to only one dimension at a time. In this experiment 84 children between the ages of 3 and 6 years were given 16 trials of a construction task, which they were either asked to do by moving two pointers simultaneously or by moving the pointers sequentially. Overall there was no main effect of condition. However, some trials were affected by condition. These results are discussed in relation to Lidster and Bremner's suggestions and Huttenlocher, Newcombe, and Sandberg's (1994) claims regarding the development of spatial understanding.     

A new approach to the formulation and testing of learning models

It is argued that current attempts to model human learning behavior commonly fail on one of two counts: either the model assumptions are artificially restricted so as to permit the application of mathematical techniques in deriving their consequences, or else the required complex assumptions are imbedded in computer programs whose technical details obscure the theoretical content of the model. The first failing is characteristic of so-called mathematical models of learning, while the second is characteristic of computer simulation models. An approach to model building which avoids both these failings is presented under the title of a black-box theory of learning. This method permits the statement of assumptions of any desired complexity in a language which clearly exhibits their theoretical content.Section II of the paper is devoted to the problem of testing and comparing alternative learning theories. The policy advocated is to abandon attempts at hypothesis testing. It is argued that, in general, we not only lack sufficient data and sufficiently powerful techniques to test hypotheses, but that the truth of a model is not really the issue of basic interest. A given model may be true in the sense that on the basis of available evidence we cannot statistically reject it, but not interesting in the sense that it provides little information about the processes underlying behavior. Rather, we should accept or reject models on the basis of how much information they provide about the way in which subjects respond to environmental structure. This attitude toward model testing is made precise by introducing a formal measure of the information content of a model. Finally, it is argued that the statistical concept of degrees-of-freedom is misleading when used in the context of model testing and should be replaced by a measure of the information absorbed from the data in estimating parameters.I would like to express my indebtedness to Ernest W. Adams and Julian Feldman for their guidance and encouragement of the research reported here.Much of the research reported here was conducted at the University of California, Berkeley, and incorporated in a Ph.D. dissertation in Logic and the Methodology of Science [15]. The research was supported in part by the Computer Institute for Social Science Research, Michigan State University.     

Chunking during serial learning by a pigeon: I. Basic evidence

Chunking by pigeons was demonstrated by comparing performance on different types of lists. Experiment 1 showed that Groups II and IV (who learned lists in which colors and achromatic geometric forms were segregated: A----B----C----D'----E' and A----B----C----D----E', respectively) executed lists more rapidly than did Group I (who learned a homogeneous list of colors: A----B----C----D----E) or Groups II and III (who learned lists consisting of unsegregated colors and forms: A----B'----C----D'----E and A----B----C'----D----E, respectively). Experiment 2 showed that Groups II and IV tolerated interruptions of the list better than did Groups I, III, and V. The accuracy of responding of Groups I, III, and V decreased as a function of the duration of the interruption and the point in the sequence at which it occurred. The performance of Group II was unaffected by interruptions; Group IV was minimally affected. These results indicate that Groups II and IV organized their lists as ordered chunks.     

Motor learning by observation: Evidence from a serial reaction time task

This study sought evidence of observational motor learning, a type of learning in which observation of the skilled performance of another person not only facilitates motor skill acquisition but does so by contributing to the formation of effector-specific motor representations. Previous research has indicated that observation of skilled performance engages cognitive processes similar to those occurring during action execution or physical practice, but has not demonstrated that these include processes involved in effector-specific representation. In two experiments, observer subjects watched the experimenter performing a serial reaction time (SRT) task with a six-item unique sequence before sequence knowledge was assessed by response time and/or free generation measures. The results suggest that: (1) subjects can acquire sequence information by watching another person performing the task (Experiments 1-2); (2) observation results in as much sequence learning as task practice when learning is measured by reaction times (RTs) and more than task practice when sequence learning is measured by free generation performance (Experiment 2, Part 1); and (3) sequence knowledge acquired by model observation can be encoded motorically--that is, in an effector-specific fashion (Experiment 2, Part 2).     

Structural complexity as a determinant of serial pattern learning

A new methodology extends the study of serial pattern learning to nonhuman organisms by constructing patterns using elements that are both familiar and motivationally meaningful to animals. Two experiments examine the ability of rats to anticipate various quantities of food as measured by running times in a runway when the quantities occur in a serial order. In Experiment 1, a serial monotonic pattern (14-7-3-1-0 food pellets) produced more rapid learning and more accurate anticipation of the various quantities (“tracking”) than a serial nonmonotonic pattern (14-1-3-7-0 food pellets), particularly with respect to the final 0-pellet element. In Experiment 2, the same monotonic pattern generated faster learning and more accurate anticipation of pattern elements than a weakly monotonic pattern (14-5-5-1-0 pellets). Associative explanations including simple excitatory or inhibitory effects, temporal anticipation, reinforcement contrast, and the number and discriminability of pairwise associations are not adequate to account for the data. Rather, the formally defined structural complexity of each pattern adequately predicts its relative difficulty.