IMPROVING FORECASTING ACCURACY IN CORPORATE PREDICTION MARKETS – A CASE STUDY IN THE AUSTRIAN MOBILE COMMUNICATION INDUSTRY

2012 ◽  
Vol 3 (3) ◽  
pp. 49-62
Author(s):  
Martin Waitz ◽  
Andreas Mild
Keyword(s):  
Predictive Ability ◽  
Market Mechanism ◽  
Prediction Markets ◽  
Data Sets ◽  
Success Rates ◽  
Prediction Market ◽  
Communication Industry ◽  
Market Participants ◽  
Market Shares ◽  
Continuous Double Auction

Corporate prediction markets forecast business issues like market shares, sales volumes or the success rates of new product developments. The improvement of its accuracy is a major topic in prediction market research. Mostly, such markets are using a continuous double auction market mechanism. We propose a method that aggregates the data provided by such a prediction market in a different way by only accounting for the most knowledgeable market participants. We demonstrate its predictive ability with a real world experiment.We want to thank Günter Fädler from pro:kons, an Austrian provider of prediction markets, for his support and providing us with the data sets used in this paper.

A Behaviorally Informed Survey-Powered Market Agent

2014 ◽  
Vol 8 (2) ◽  
pp. 1-28
Author(s):  
Jessica Inchauspe ◽  
Pavel Atanasov ◽  
Barbara Mellers ◽  
Philip Tetlock ◽  
Lyle Ungar
Keyword(s):  
Prediction Markets ◽  
Opinion Poll ◽  
Large Set ◽  
Prediction Market ◽  
Probability Estimates ◽  
Individual Probability ◽  
Continuous Double Auction ◽  
Zero Intelligence ◽  
Poll Data ◽  
Independent Probability

We introduce a new method for converting individual probability estimates (obtained through surveys) into market orders for use in a Continuous Double Auction prediction market. Our Survey-Powered Market Agent (SPMA) algorithm is based on actual forecaster behavior, and offers notable advantages over existing market agent algorithms such as Zero Intelligence Plus (ZIP) agents: SPMAs only require probability estimates (and not bid direction nor quantity), are more behaviorally realistic, and work well when probabilities change over time. We validate SPMA using prediction market data and probability estimates elicited through surveys from a large set of forecasters on 88 individual forecasting problems over the course of a year. SPMA outperforms simple averages of the same probability forecasts and is competitive with sophisticated opinion poll aggregation methods and prediction markets. We use a rich set of market and poll data to empirically test the assumptions behind SPMA’s operation. In addition to aggregation efficiency, SPMA provides a framework for studying how forecasters convert probability estimates into trading orders, and offers a foundation for building hybrid markets which mix market traders and individuals producing independent probability estimates.

scholarly journals Pathway-guided deep neural network toward interpretable and predictive modeling of drug sensitivity

2020 ◽  
Author(s):  
Lei Deng ◽  
Yideng Cai ◽  
Wenhao Zhang ◽  
Wenyi Yang ◽  
Bo Gao ◽  
...  
Keyword(s):  
Neural Network ◽  
Cancer Cells ◽  
Drug Targets ◽  
Deep Neural Network ◽  
Drug Sensitivity ◽  
Predictive Ability ◽  
Biological Knowledge ◽  
Data Sets ◽  
Source Codes ◽  
Sensitivity Data

AbstractMotivationTo efficiently save cost and reduce risk in drug research and development, there is a pressing demand to develop in-silico methods to predict drug sensitivity to cancer cells. With the exponentially increasing number of multi-omics data derived from high-throughput techniques, machine learning-based methods have been applied to the prediction of drug sensitivities. However, these methods have drawbacks either in the interpretability of mechanism of drug action or limited performance in modeling drug sensitivity.ResultsIn this paper, we presented a pathway-guided deep neural network model, referred to as pathDNN, to predict the drug sensitivity to cancer cells. Biological pathways describe a group of molecules in a cell that collaborates to control various biological functions like cell proliferation and death, thereby abnormal function of pathways can result in disease. To make advantage of both the excellent predictive ability of deep neural network and the biological knowledge of pathways, we reshape the canonical DNN structure by incorporating a layer of pathway nodes and their connections to input gene nodes, which makes the DNN model more interpretable and predictive compared to canonical DNN. We have conducted extensive performance evaluations on multiple independent drug sensitivity data sets, and demonstrate that pathDNN significantly outperformed canonical DNN model and seven other classical regression models. Most importantly, we observed remarkable activity decreases of disease-related pathway nodes during forward propagation upon inputs of drug targets, which implicitly corresponds to the inhibition effect of disease-related pathways induced by drug treatment on cancer cells. Our empirical experiments show that pathDNN achieves pharmacological interpretability and predictive ability in modeling drug sensitivity to cancer cells.AvailabilityThe web server, as well as the processed data sets and source codes for reproducing our work, is available at http://pathdnn.denglab.org

scholarly journals Empirics of Korean Shipping Companies’ Default Predictions

Risks ◽  
2021 ◽  
Vol 9 (9) ◽  
pp. 159
Author(s):  
Sunghwa Park ◽  
Hyunsok Kim ◽  
Janghan Kwon ◽  
Taeil Kim
Keyword(s):  
Logit Model ◽  
Prediction Models ◽  
Predictive Ability ◽  
Financial Ratios ◽  
Probability Of Default ◽  
Shipping Industry ◽  
Market Participants ◽  
Korean Economy ◽  
Default Prediction ◽  
Financial Characteristics

In this paper, we use a logit model to predict the probability of default for Korean shipping companies. We explore numerous financial ratios to find predictors of a shipping firm’s failure and construct four default prediction models. The results suggest that a model with industry specific indicators outperforms other models in predictive ability. This finding indicates that utilizing information about unique financial characteristics of the shipping industry may enhance the performance of default prediction models. Given the importance of the shipping industry in the Korean economy, this study can benefit both policymakers and market participants.

scholarly journals Factors determining generalization in deep learning models for scoring COVID-CT images

2021 ◽  
Vol 18 (6) ◽  
pp. 9264-9293
Author(s):  
Michael James Horry ◽  
◽  
Subrata Chakraborty ◽  
Biswajeet Pradhan ◽  
Maryam Fallahpoor ◽  
...  
Keyword(s):  
Deep Learning ◽  
Predictive Accuracy ◽  
Gabor Filter ◽  
Predictive Ability ◽  
Histogram Equalization ◽  
Data Sets ◽  
Lung Involvement ◽  
Learning Models ◽  
Key Factors ◽  
Independent Dataset

<abstract> <p>The COVID-19 pandemic has inspired unprecedented data collection and computer vision modelling efforts worldwide, focused on the diagnosis of COVID-19 from medical images. However, these models have found limited, if any, clinical application due in part to unproven generalization to data sets beyond their source training corpus. This study investigates the generalizability of deep learning models using publicly available COVID-19 Computed Tomography data through cross dataset validation. The predictive ability of these models for COVID-19 severity is assessed using an independent dataset that is stratified for COVID-19 lung involvement. Each inter-dataset study is performed using histogram equalization, and contrast limited adaptive histogram equalization with and without a learning Gabor filter. We show that under certain conditions, deep learning models can generalize well to an external dataset with F1 scores up to 86%. The best performing model shows predictive accuracy of between 75% and 96% for lung involvement scoring against an external expertly stratified dataset. From these results we identify key factors promoting deep learning generalization, being primarily the uniform acquisition of training images, and secondly diversity in CT slice position.</p> </abstract>

scholarly journals Prediction Markets for Crowdsourcing

2018 ◽  
Author(s):  
Christian Horn ◽  
Marcel Bogers ◽  
Alexander Brem*
Keyword(s):  
Stock Market ◽  
Open Innovation ◽  
Innovation Process ◽  
Business Practice ◽  
Prediction Markets ◽  
Knowledge Sources ◽  
Prediction Market ◽  
Online Platform ◽  
External Knowledge ◽  
Opening Up

Crowdsourcing is an increasingly important phenomenon that is fundamentally changing how companies create and capture value. There are still important questions with respect to how crowdsourcing works and can be applied in practice, especially in business practice. In this chapter, we focus on prediction markets as a mechanism and tool to tap into a crowd in the early stages of an innovation process. The act of opening up to external knowledge sources is also in line with the growing interest in open innovation. One example of a prediction market, a virtual stock market, is applied to open innovation through an online platform. We show that use of mechanisms of internal crowdsourcing with prediction markets can outperform use of external crowds.

scholarly journals Monte Carlo and Reconstruction Membership Inference Attacks against Generative Models

2019 ◽  
Vol 2019 (4) ◽  
pp. 232-249 ◽  
Author(s):  
Benjamin Hilprecht ◽  
Martin Härterich ◽  
Daniel Bernau
Keyword(s):  
Data Privacy ◽  
Information Leakage ◽  
Generative Models ◽  
Generative Model ◽  
Training Data ◽  
Generative Adversarial Networks ◽  
Data Sets ◽  
Success Rates ◽  
Model Quality ◽  
Type Formalization

Abstract We present two information leakage attacks that outperform previous work on membership inference against generative models. The first attack allows membership inference without assumptions on the type of the generative model. Contrary to previous evaluation metrics for generative models, like Kernel Density Estimation, it only considers samples of the model which are close to training data records. The second attack specifically targets Variational Autoencoders, achieving high membership inference accuracy. Furthermore, previous work mostly considers membership inference adversaries who perform single record membership inference. We argue for considering regulatory actors who perform set membership inference to identify the use of specific datasets for training. The attacks are evaluated on two generative model architectures, Generative Adversarial Networks (GANs) and Variational Autoen-coders (VAEs), trained on standard image datasets. Our results show that the two attacks yield success rates superior to previous work on most data sets while at the same time having only very mild assumptions. We envision the two attacks in combination with the membership inference attack type formalization as especially useful. For example, to enforce data privacy standards and automatically assessing model quality in machine learning as a service setups. In practice, our work motivates the use of GANs since they prove less vulnerable against information leakage attacks while producing detailed samples.

scholarly journals Avian succession along ecological gradients: Insight from species-poor and species-rich communities of Sylvia warblers

2011 ◽  
Vol 57 (3) ◽  
pp. 307-317 ◽  
Author(s):  
Alban Guillaumet ◽  
Roger Prodon
Keyword(s):  
Interspecific Competition ◽  
Species Turnover ◽  
Predictive Ability ◽  
Distribution Model ◽  
Joint Analysis ◽  
Data Sets ◽  
Ecological Gradients ◽  
Distribution Models ◽  
Ecological Release ◽  
Low Diversity

Abstract The mechanisms responsible for species replacement during ecological successions is a long-standing and open debate. In this study, we examined the distribution of the Sardinian warbler Sylvia melanocephala along two grassland-to-forest gradients, one in a high-diversity area (Albera-Aspres chain in Catalonia: eight Sylvia warbler species) and one in a low-diversity area (Mount Hymittos in Greece: four species). In Catalonia, distribution models suggested that the apparent exclusion of S. melanocephala from the open and forest ends of the gradient may be explained entirely by the preference of S. melanocephala for mid-successional shrublands. However, a joint analysis of both data sets revealed that: 1) S. melanocephala was more evenly distributed along the vegetation gradient in Greece, suggesting ecological release in the low-diversity area; and 2) a distribution model assuming interspecific competition (based on the distribution of Sylvia species showing a negative co-occurrence pattern with S. melanocephala) had a significantly higher predictive ability than a distribution model based on habitat variables alone. Our study supports the view that species turnover along ecological gradients generally results from a combination of intrinsic preferences and interspecific competition.

Accruals and Future Stock Returns: Tests of the Naïve Investor Hypothesis

2000 ◽  
Vol 15 (2) ◽  
pp. 161-181 ◽  
Author(s):  
Ashiq Ali ◽  
Lee-Seok Hwang ◽  
Mark A. Trombley
Keyword(s):  
Transaction Costs ◽  
Stock Returns ◽  
Stock Price ◽  
Predictive Ability ◽  
Relevant Information ◽  
Relative Magnitude ◽  
Quarterly Earnings ◽  
Market Participants ◽  
Flow Components ◽  
Reported Earnings

We explore whether the association between accruals and future returns documented by Sloan (1996) is due to fixation by naïve investors on the total amount of reported earnings without regard for the relative magnitude of the accrual and cash flow components. Contrary to the predictions of the naïve investor hypothesis, we find that the predictive ability of accruals for subsequent annual returns and for quarterly earnings announcement stock returns is not lower for large firms or for firms followed more by analysts or held more by institutions. Further, we find that the ability of accruals to predict future returns does not seem to depend on stock price or transaction volume, measures of transaction costs, also contrary to predictions of the naïve investor hypothesis. These results are robust to regression and hedge portfolio tests. We conclude that the predictive ability of accruals for subsequent returns does not seem to be due to the inability of market participants to understand value-relevant information.

Development of disaggregate mode choice models of intercity travel in Canada

1992 ◽  
Vol 19 (6) ◽  
pp. 965-974 ◽  
Author(s):  
Walid M. Abdelwahab ◽  
J. David Innes ◽  
Albert M. Stevens
Keyword(s):  
Travel Behavior ◽  
Mode Choice ◽  
Geographical Location ◽  
Predictive Ability ◽  
Multinomial Logit ◽  
Choice Models ◽  
Trip Purpose ◽  
Market Shares ◽  
Intercity Travel ◽  
Statistics Canada

This paper reports and discusses the results of an effort to develop disaggregate behavioral mode choice models of intercity travel in Canada. Currently available data bases of intercity travel in Canada are reviewed. The feasibility of using data from national travel surveys to develop statistically reliable intercity mode choice models is examined, and directions for future disaggregate data collection efforts are offered. The models developed are of the multinomial logit (MNL) type which included all intercity passenger travel modes: auto, air, bus, and rail. For purposes of estimation, the travel market was segmented by trip length (short, long); trip purpose (business, recreational); and geographical location of the trip (east, west). Then, a separate model was estimated in each sector. The models were estimated using the data collected by Statistics Canada as a part of the Labor Force Survey (The Canadian Travel Survey, CTS). The quality of the calibrated models varied from one region to another and from one travel sector to another. Overall, the models were reasonably accurate in predicting modal shares of the most frequently used modes (auto and air). The underrepresentation of the bus and rail modes in the data sets led to a deterioration in the performance of the models in predicting market shares of these two modes. More specifically, the predictive ability of the models measured by the likelihood ratio index varied from a low of 0.58 in the short travel sector to a high of 0.94 in the long travel sector. The transferability of the models described in this study was recently examined by Abdelwahab (1991). Key words: mode choice, disaggregate, travel behavior, multinomial logit, intercity, data base.

Sign in / Sign up

Export Citation Format

Share Document