|
on Big Data |
| By: | Esther Rolf; Jonathan Proctor; Tamma Carleton; Ian Bolliger; Vaishaal Shankar; Miyabi Ishihara; Benjamin Recht; Solomon Hsiang |
| Abstract: | Combining satellite imagery with machine learning (SIML) has the potential to address global challenges by remotely estimating socioeconomic and environmental conditions in data-poor regions, yet the resource requirements of SIML limit its accessibility and use. We show that a single encoding of satellite imagery can generalize across diverse prediction tasks (e.g. forest cover, house price, road length). Our method achieves accuracy competitive with deep neural networks at orders of magnitude lower computational cost, scales globally, delivers label super-resolution predictions, and facilitates characterizations of uncertainty. Since image encodings are shared across tasks, they can be centrally computed and distributed to unlimited researchers, who need only fit a linear regression to their own ground truth data in order to achieve state-of-the-art SIML performance. |
| JEL: | C02 C8 O13 O18 Q5 R1 |
| Date: | 2020–11 |
| URL: | http://d.repec.org/n?u=RePEc:nbr:nberwo:28045&r=all |
| By: | Steven Lehrer; Tian Xie (Queen's University) |
| Abstract: | There exists significant hype regarding how much machine learning and incorporating social media data can improve forecast accuracy in commercial applications. To assess if the hype is warranted, we use data from the film industry in simulation experiments that contrast econometric approaches with tools from the predictive analytics literature. Further, we propose new strategies that combine elements from each literature in a bid to capture richer patterns of heterogeneity in the underlying relationship governing revenue. Our results demonstrate the importance of social media data and value from hybrid strategies that combine econometrics and machine learning when conducting forecasts with new big data sources. Specifically, while both least squares support vector regression and recursive partitioning strategies greatly outperform dimension reduction strategies and traditional econometrics approaches in forecast accuracy, there are further significant gains from using hybrid approaches. Further, Monte Carlo experiments demonstrate that these benefits arise from the significant heterogeneity in how social media measures and other film characteristics influence box office outcomes. |
| Keywords: | Machine Learning, Model Specification, Heteroskedasticity, Heterogeneity, Social Media, Big Data |
| JEL: | C52 L82 D03 M21 C53 |
| Date: | 2020–10 |
| URL: | http://d.repec.org/n?u=RePEc:qed:wpaper:1449&r=all |
| By: | Maria Concetta Ambra (Department of Social Sciences and Economics, Sapienza University of Rome) |
| Abstract: | This article focuses on Amazon Mechanical Turk (AMT), the crowdsourcing platform created by Amazon, with the aim to enrich our knowledge of this specific platform and to contribute to the debate on ‘platform economy’. In light of the massive changes triggered by the new digital revolution, many scholars have recently examined how platform work has changed, by exploring transformations in employee status and the new content of platform work. This article addresses two interrelated questions: to what extent and in what ways does AMT chal-lenge the boundaries between paid and unpaid digital labour? How does AMT exploit online labour to extract surplus value? The research was undertaken from December 2018 and July 2019, through the collection of 50 doc-uments originating from three Amazon web sites. These documents have been examined though the technique of content analysis by adopting the NVivo software. In conclusion, it explains how Amazon has been able to develop a hybrid system of human-machine work. This specific model can be also fruitful used to speed up the machine learning process and to make it more accurate. |
| Keywords: | Amazon Mechanical Turk; Crowdsourcing Platform; Digital Piecework; Intellectual Property Rights; Machine Learning |
| JEL: | J30 J83 D20 O30 |
| Date: | 2020–11 |
| URL: | http://d.repec.org/n?u=RePEc:saq:wpaper:19/20&r=all |
| By: | Brian Quistorff; Gentry Johnson |
| Abstract: | Restricting randomization in the design of experiments (e.g., using blocking/stratification, pair-wise matching, or rerandomization) can improve the treatment-control balance on important covariates and therefore improve the estimation of the treatment effect, particularly for small- and medium-sized experiments. Existing guidance on how to identify these variables and implement the restrictions is incomplete and conflicting. We identify that differences are mainly due to the fact that what is important in the pre-treatment data may not translate to the post-treatment data. We highlight settings where there is sufficient data to provide clear guidance and outline improved methods to mostly automate the process using modern machine learning (ML) techniques. We show in simulations using real-world data, that these methods reduce both the mean squared error of the estimate (14%-34%) and the size of the standard error (6%-16%). |
| Date: | 2020–10 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2010.15966&r=all |
| By: | Tadeu A. Ferreira |
| Abstract: | Inspired by the developments in deep generative models, we propose a model-based RL approach, coined Reinforced Deep Markov Model (RDMM), designed to integrate desirable properties of a reinforcement learning algorithm acting as an automatic trading system. The network architecture allows for the possibility that market dynamics are partially visible and are potentially modified by the agent's actions. The RDMM filters incomplete and noisy data, to create better-behaved input data for RL planning. The policy search optimisation also properly accounts for state uncertainty. Due to the complexity of the RKDF model architecture, we performed ablation studies to understand the contributions of individual components of the approach better. To test the financial performance of the RDMM we implement policies using variants of Q-Learning, DynaQ-ARIMA and DynaQ-LSTM algorithms. The experiments show that the RDMM is data-efficient and provides financial gains compared to the benchmarks in the optimal execution problem. The performance improvement becomes more pronounced when price dynamics are more complex, and this has been demonstrated using real data sets from the limit order book of Facebook, Intel, Vodafone and Microsoft. |
| Date: | 2020–11 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2011.04391&r=all |
| By: | Pooja Gupta; Angshul Majumdar; Emilie Chouzenoux; Giovanni Chierchia |
| Abstract: | This work proposes a supervised multi-channel time-series learning framework for financial stock trading. Although many deep learning models have recently been proposed in this domain, most of them treat the stock trading time-series data as 2-D image data, whereas its true nature is 1-D time-series data. Since the stock trading systems are multi-channel data, many existing techniques treating them as 1-D time-series data are not suggestive of any technique to effectively fusion the information carried by the multiple channels. To contribute towards both of these shortcomings, we propose an end-to-end supervised learning framework inspired by the previously established (unsupervised) convolution transform learning framework. Our approach consists of processing the data channels through separate 1-D convolution layers, then fusing the outputs with a series of fully-connected layers, and finally applying a softmax classification layer. The peculiarity of our framework - SuperDeConFuse (SDCF), is that we remove the nonlinear activation located between the multi-channel convolution layers and the fully-connected layers, as well as the one located between the latter and the output layer. We compensate for this removal by introducing a suitable regularization on the aforementioned layer outputs and filters during the training phase. Specifically, we apply a logarithm determinant regularization on the layer filters to break symmetry and force diversity in the learnt transforms, whereas we enforce the non-negativity constraint on the layer outputs to mitigate the issue of dead neurons. This results in the effective learning of a richer set of features and filters with respect to a standard convolutional neural network. Numerical experiments confirm that the proposed model yields considerably better results than state-of-the-art deep learning techniques for real-world problem of stock trading. |
| Date: | 2020–11 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2011.04364&r=all |
| By: | Hal Ashton |
| Abstract: | Campbell-Goodhart's law relates to the causal inference error whereby decision-making agents aim to influence variables which are correlated to their goal objective but do not reliably cause it. This is a well known error in Economics and Political Science but not widely labelled in Artificial Intelligence research. Through a simple example, we show how off-the-shelf deep Reinforcement Learning (RL) algorithms are not necessarily immune to this cognitive error. The off-policy learning method is tricked, whilst the on-policy method is not. The practical implication is that naive application of RL to complex real life problems can result in the same types of policy errors that humans make. Great care should be taken around understanding the causal model that underpins a solution derived from Reinforcement Learning. |
| Date: | 2020–11 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2011.01010&r=all |
| By: | Alexander Wong; Andrew Hryniowski; Xiao Yu Wang |
| Abstract: | The success of deep learning in recent years have led to a significant increase in interest and prevalence for its adoption to tackle financial services tasks. One particular question that often arises as a barrier to adopting deep learning for financial services is whether the developed financial deep learning models are fair in their predictions, particularly in light of strong governance and regulatory compliance requirements in the financial services industry. A fundamental aspect of fairness that has not been explored in financial deep learning is the concept of trust, whose variations may point to an egocentric view of fairness and thus provide insights into the fairness of models. In this study we explore the feasibility and utility of a multi-scale trust quantification strategy to gain insights into the fairness of a financial deep learning model, particularly under different scenarios at different scales. More specifically, we conduct multi-scale trust quantification on a deep neural network for the purpose of credit card default prediction to study: 1) the overall trustworthiness of the model 2) the trust level under all possible prediction-truth relationships, 3) the trust level across the spectrum of possible predictions, 4) the trust level across different demographic groups (e.g., age, gender, and education), and 5) distribution of overall trust for an individual prediction scenario. The insights for this proof-of-concept study demonstrate that such a multi-scale trust quantification strategy may be helpful for data scientists and regulators in financial services as part of the verification and certification of financial deep learning solutions to gain insights into fairness and trust of these solutions. |
| Date: | 2020–11 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2011.01961&r=all |
| By: | Olivier Guéant (CES - Centre d'économie de la Sorbonne - UP1 - Université Panthéon-Sorbonne - CNRS - Centre National de la Recherche Scientifique); Iuliia Manziuk (CES - Centre d'économie de la Sorbonne - UP1 - Université Panthéon-Sorbonne - CNRS - Centre National de la Recherche Scientifique); Jiang Pu |
| Abstract: | When firms want to buy back their own shares, they have a choice between several alternatives. If they often carry out open market repurchase, they also increasingly rely on banks through complex buyback contracts involving option components, e.g. accelerated share repurchase contracts, VWAP-minus profit-sharing contracts, etc. The entanglement between the execution problem and the option hedging problem makes the management of these contracts a difficult task that should not boil down to simple Greek-based risk hedging, contrary to what happens with classical books of options. In this paper, we propose a machine learning method to optimally manage several types of buyback contract. In particular, we recover strategies similar to those obtained in the literature with partial differential equation and recombinant tree methods and show that our new method, which does not suffer from the curse of dimensionality, enables to address types of contract that could not be addressed with grid or tree methods. |
| Keywords: | ASR contracts,Optimal stopping,Stochastic optimal control,Deep learning,Recurrent neural networks,Reinforcement learning |
| Date: | 2020–11–04 |
| URL: | http://d.repec.org/n?u=RePEc:hal:wpaper:hal-02987889&r=all |
| By: | Olivier Guéant (CES - Centre d'économie de la Sorbonne - UP1 - Université Panthéon-Sorbonne - CNRS - Centre National de la Recherche Scientifique); Iuliia Manziuk (CES - Centre d'économie de la Sorbonne - UP1 - Université Panthéon-Sorbonne - CNRS - Centre National de la Recherche Scientifique); Jiang Pu |
| Abstract: | When firms want to buy back their own shares, they have a choice between several alternatives. If they often carry out open market repurchase, they also increasingly rely on banks through complex buyback contracts involving option components, e.g. accelerated share repurchase contracts, VWAP-minus profit-sharing contracts, etc. The entanglement between the execution problem and the option hedging problem makes the management of these contracts a difficult task that should not boil down to simple Greek-based risk hedging, contrary to what happens with classical books of options. In this paper, we propose a machine learning method to optimally manage several types of buyback contract. In particular, we recover strategies similar to those obtained in the literature with partial differential equation and recombinant tree methods and show that our new method, which does not suffer from the curse of dimensionality, enables to address types of contract that could not be addressed with grid or tree methods. |
| Keywords: | ASR contracts,Optimal stopping,Stochastic optimal control,Deep learning,Recurrent neural networks,Reinforcement learning |
| Date: | 2020–11–04 |
| URL: | http://d.repec.org/n?u=RePEc:hal:cesptp:hal-02987889&r=all |
| By: | Sinha, Pankaj; Verma, Aniket; Shah, Purav; Singh, Jahnavi; Panwar, Utkarsh |
| Abstract: | This paper aims at determining the various economic and non-economic factors that can influence the voting behaviour in the forthcoming United States Presidential Election using Lasso regression, a Machine learning algorithm. Even though contemporary discussions on the subject of the United States Presidential Election suggest that the level of unemployment in the economy will be a significant factor in determining the result of the election, in our study, it has been found that the rate of unemployment will not be the only significant factor in forecasting the election. However, various other economic factors such as the inflation rate, rate of economic growth, and exchange rates will not have a significant influence on the election result. The June Gallup Rating, is not the only significant factor for determining the result of the forthcoming presidential election. In addition to the June Gallup Rating, various other non-economic factors such as the performance of the contesting political parties in the midterm elections, Campaign spending by the contesting parties and scandals of the Incumbent President will also play a significant role in determining the result of the forthcoming United States Presidential Election. The paper explores the influence of all the aforementioned economic and non-economic factors on the voting behaviour of the voters in the forthcoming United States Presidential Election. The proposed Lasso Regression model forecasts that the vote share for the incumbent Republican Party to be 41.63% in the 2020 US presidential election. This means that the incumbent party is most likely to lose the upcoming election. |
| Keywords: | US Presidential Election, Machine Learning, Lasso Regression, Economic Factors, None Economic Factor, Forecasting, Prediction |
| JEL: | C10 C13 C15 C6 C61 C63 C8 |
| Date: | 2020–10–13 |
| URL: | http://d.repec.org/n?u=RePEc:pra:mprapa:103889&r=all |
| By: | Diunugala, Hemantha Premakumara; Mombeuil, Claudel |
| Abstract: | Purpose: This study compares three different methods to predict foreign tourist arrivals (FTAs) to Sri Lanka from top-ten countries and also attempts to find the best-fitted forecasting model for each country using five model performance evaluation criteria. Methods: This study employs two different univariate-time-series approaches and one Artificial Intelligence (AI) approach to develop models that best explain the tourist arrivals to Sri Lanka from the top-ten tourist generating countries. The univariate-time series approach contains two main types of statistical models, namely Deterministic Models and Stochastic Models. Results: The results show that Winter’s exponential smoothing and ARIMA are the best methods to forecast tourist arrivals to Sri Lanka. Furthermore, the results show that the accuracy of the best forecasting model based on MAPE criteria for the models of India, China, Germany, Russia, and Australia fall between 5 to 9 percent, whereas the accuracy levels of models for the UK, France, USA, Japan, and the Maldives fall between 10 to 15 percent. Implications: The overall results of this study provide valuable insights into tourism management and policy development for Sri Lanka. Successful forecasting of FTAs for each market source provide a practical planning tool to destination decision-makers. |
| Keywords: | foreign tourist arrivals, winter’s exponential smoothing, ARIMA, simple recurrent neural network, Sri Lanka |
| JEL: | C45 C5 Z0 |
| Date: | 2020–10–30 |
| URL: | http://d.repec.org/n?u=RePEc:pra:mprapa:103779&r=all |
| By: | Jianqing Fan; Ricardo P. Masini; Marcelo C. Medeiros |
| Abstract: | The measurement of treatment (intervention) effects on a single (or just a few) treated unit(s) based on counterfactuals constructed from artificial controls has become a popular practice in applied statistics and economics since the proposal of the synthetic control method. In high-dimensional setting, we often use principal component or (weakly) sparse regression to estimate counterfactuals. Do we use enough data information? To better estimate the effects of price changes on the sales in our case study, we propose a general framework on counterfactual analysis for high dimensional dependent data. The framework includes both principal component regression and sparse linear regression as specific cases. It uses both factor and idiosyncratic components as predictors for improved counterfactual analysis, resulting a method called Factor-Adjusted Regularized Method for Treatment (FarmTreat) evaluation. We demonstrate convincingly that using either factors or sparse regression is inadequate for counterfactual analysis in many applications and the case for information gain can be made through the use of idiosyncratic components. We also develop theory and methods to formally answer the question if common factors are adequate for estimating counterfactuals. Furthermore, we consider a simple resampling approach to conduct inference on the treatment effect as well as bootstrap test to access the relevance of the idiosyncratic components. We apply the proposed method to evaluate the effects of price changes on the sales of a set of products based on a novel large panel of sale data from a major retail chain in Brazil and demonstrate the benefits of using additional idiosyncratic components in the treatment effect evaluations. |
| Date: | 2020–11 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2011.03996&r=all |
| By: | Gries, Thomas; Naude, Wim |
| Abstract: | The economic impact of Artificial Intelligence (AI) is studied using a (semi) endogenous growth model with two novel features. First, the task approach from labor economics is reformulated and integrated into a growth model. Second, the standard represen- tative household assumption is rejected, so that aggregate demand restrictions can be introduced. With these novel features it is shown that (i) AI automation can decrease the share of labor income no matter the size of the elasticity of substitution between AI and labor, and (ii) when this elasticity is high, AI will unambiguously reduce aggre- gate demand and slow down GDP growth, even in the face of the positive technology shock that AI entails. If the elasticity of substitution is low, then GDP, productivity and wage growth may however still slow down, because the economy will then fail to benefit from the supply-side driven capacity expansion potential that AI can deliver. The model can thus explain why advanced countries tend to experience, despite much AI hype, the simultaneous existence of rather high employment with stagnating wages, productivity, and GDP. |
| Keywords: | Technology,artificial intelligence,productivity,labor demand,income distribution,growth theory |
| JEL: | O47 O33 J24 E21 E25 |
| Date: | 2020 |
| URL: | http://d.repec.org/n?u=RePEc:zbw:vfsc20:224623&r=all |
| By: | Loann Desboulets (AMSE - Aix-Marseille Sciences Economiques - EHESS - École des hautes études en sciences sociales - AMU - Aix Marseille Université - ECM - École Centrale de Marseille - CNRS - Centre National de la Recherche Scientifique) |
| Abstract: | This paper is devoted to practical use of the Manifold Selection method presented in Desboulets (2020). In a first part, I present an application on financial data. The data I use are continuous futures contracts underlying commodities. These are multivariate time series, for the period 1985-2020. Representing correlations in financial data as graphs is a common task, useful in Finance for risk assessment. However, these graphs are often too complex, and involve many small connections. Therefore, the graphs can be simplified using variable selection, to remove these small correlations. Here, I use Manifold Selection to build sparse graphical models. Non-linear manifolds can represent interconnected markets where the major drivers of prices are unobserved. The results indicate the market is more strongly interconnected when using non-linear manifold selection than when using linear graphical models. I also propose a new method for filling missing values in time series data. I run a simulation and show that the method performs well in case of several consecutive missing values. |
| Keywords: | Non-parametric,Non-linear Manifolds,Variable Selection,Neural Networks |
| Date: | 2020–11–03 |
| URL: | http://d.repec.org/n?u=RePEc:hal:wpaper:hal-02986982&r=all |
| By: | Qing Yang; Zhenning Hong; Ruyan Tian; Tingting Ye; Liangliang Zhang |
| Abstract: | In this paper, we document a novel machine learning based bottom-up approach for static and dynamic portfolio optimization on, potentially, a large number of assets. The methodology overcomes many major difficulties arising in current optimization schemes. For example, we no longer need to compute the covariance matrix and its inverse for mean-variance optimization, therefore the method is immune from the estimation error on this quantity. Moreover, no explicit calls of optimization routines are needed. Applications to a bottom-up mean-variance-skewness-kurtosis or CRRA (Constant Relative Risk Aversion) optimization with short-sale portfolio constraints in both simulation and real market (China A-shares and U.S. equity markets) environments are studied and shown to perform very well. |
| Date: | 2020–11 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2011.00572&r=all |
| By: | Ferman, Bruno; Lima, Lycia; Riva, Flavio |
| Abstract: | This paper investigates how technologies that use different combinations of artificial and human intelligence are incorporated into classroom instruction, and how they ultimately affect students' outcomes. We conducted a field experiment to study two technologies that allow teachers to outsource grading and feedback tasks on writing practices. The first technology is a fully automated evaluation system that provides instantaneous scores and feedback. The second one uses human graders as an additional resource to enhance grading and feedback quality in aspects in which the automated system arguably falls short. Both technologies significantly improved students' essay scores, and the additional inputs from human graders did not improve effectiveness. Furthermore, the technologies similarly helped teachers engage more frequently on nonroutine tasks that supported the individualization of pedagogy. Our results are informative about the potential of artificial intelligence to expand the set of tasks that can be automated, and on how advances in artificial intelligence may relocate human labor to tasks that remain out of reach of automation. |
| Keywords: | artificial intelligence; technological change; automated writing evaluation; routine and nonroutine tasks |
| JEL: | I21 I22 I25 |
| Date: | 2020–11–04 |
| URL: | http://d.repec.org/n?u=RePEc:pra:mprapa:103934&r=all |