Stock market forecasting with super-high dimensional time-series data using ConvLSTM, trend sampling, and specialized data augmentation

Lee, Si Woon; Kim, Ha Young

doi:10.1016/j.eswa.2020.113704

Cited by 85 publications

(36 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Li et al [10] proposed a unique approach to text-data fusion that was based on LSTM's model, using four different sentiment dictionaries for a five-year stock market prediction experiment on stock technical indicators and text news data from Hong Kong. However, Lee and Kim [11] proposed a capsule network model based on the transformer encoder (CapTE), which uses the transformer encoder to extract the deep semantic features of social media texts, and then capture the structural relationships of these texts through the capsule network. In addition, attention mechanisms have been applied in many fields since their emergence, and they have now been applied to financial market research problems, especially the temporal attention mechanism, to adapt to the time series of financial data and obtain critical information.…”

Section: Related Workmentioning

confidence: 99%

Incorporating Transformers and Attention Networks for Stock Movement Prediction

Liu

et al. 2022

Complexity

View full text Add to dashboard Cite

Predicting stock movements is a valuable research field that can help investors earn more profits. As with time-series data, the stock market is time-dependent and the value of historical information may decrease over time. Accurate prediction can be achieved by mining valuable information with words on social platforms and further integrating it with actual stock market conditions. However, many methods still cannot effectively dig deep into hidden information, integrate text and stock prices, and ignore the temporal dependence. Therefore, to solve the above problems, we propose a transformer-based attention network framework that uses historical text and stock prices to capture the temporal dependence of financial data. Among them, the transformer model and attention mechanism are used for feature extraction of financial data, which has fewer applications in the financial field, and effective analysis of key information to achieve an accurate prediction. A large number of experiments have proved the effectiveness of our proposed method. The actual simulation experiment verifies that our model has practical application value.

show abstract

Section: Related Workmentioning

confidence: 99%

Incorporating Transformers and Attention Networks for Stock Movement Prediction

Liu

et al. 2022

Complexity

View full text Add to dashboard Cite

show abstract

“…Since the main target of our proposal is Big Data scenarios, we have focused our experimentation on all the available variables to process as much data as possible. The study in [34] used the following models: Decision Tree (DT), Random Forest (RF), AdaBoost (AB), Linear Discriminant Analysis (LDA), 3 and K-Nearest Neighbors (KNN) with k = 9. For the (DT, RF, AB) models, the referred to work set the minimum number of samples required to split a node to 20, and they set the number of base estimators to 100 for (RF and AB).…”

Section: Modelsmentioning

confidence: 99%

“…With the expansion of new technologies, the volume of data generated is growing by leaps and bounds. Until now, the typical time series data comes from well-known fields, for example, from the stock market [3], from industry with power consumption logs [4], or from medical fields with specific applications, such as electrocardiograms [5]. However, nowadays we have access to a lot of new devices like smartwatches which continuously generate information through their incorporated sensors, such as heart rate, temperature or humidity monitors.…”

Section: Introductionmentioning

confidence: 99%

SCMFTS: Scalable and Distributed Complexity Measures and Features for Univariate and Multivariate Time Series in Big Data Environments

Baldan

Peralta

Saeys

et al. 2021

Int J Comput Intell Syst

View full text Add to dashboard Cite

Time series data are becoming increasingly important due to the interconnectedness of the world. Classical problems, which are getting bigger and bigger, require more and more resources for their processing, and Big Data technologies offer many solutions. Although the principal algorithms for traditional vector-based problems are available in Big Data environments, the lack of tools for time series processing in these environments needs to be addressed. In this work, we propose a scalable and distributed time series transformation for Big Data environments based on well-known time series features (SCMFTS), which allows practitioners to apply traditional vector-based algorithms to time series problems. The proposed transformation, along with the algorithms available in Spark, improved the best results in the state-of-the-art on the Wearable Stress and Affect Detection dataset, which is the biggest publicly available multivariate time series dataset in the University of California Irvine (UCI) Machine Learning Repository. In addition, SCMFTS showed a linear relationship between its runtime and the number of processed time series, demonstrating a linear scalable behavior, which is mandatory in Big Data environments. SCMFTS has been implemented in the Scala programming language for the Apache Spark framework, and the code is publicly available.

show abstract

“…Small size of training images P a may lead to overfitting, to which a powerful solution is data augmentation (Lee & Kim, 2020) that creates fake training images which will be included to the training set P a . We choose data augmentation due to its ease to implement.…”

Section: Datasetmentioning

confidence: 99%

Fruit category classification by fractional Fourier entropy with rotation angle vector grid and stacked sparse autoencoder

2021

View full text Add to dashboard Cite

Aim Fruit category classification is important in factory packing and transportation, price prediction, dietary intake, and so forth. Methods This study proposed a novel artificial intelligence system to classify fruit categories. First, 2D fractional Fourier entropy with rotation angle vector grid was used to extract features from fruit images. Afterwards, a five‐layer stacked sparse autoencoder was used as the classifier. Results Ten runs on the test set showed our method achieved a micro‐averaged F1 score of 95.08% for an 18‐category fruit dataset. Conclusion Our method gives better micro‐averaged F1 score than 10 state‐of‐the‐art approaches.

show abstract

Stock market forecasting with super-high dimensional time-series data using ConvLSTM, trend sampling, and specialized data augmentation

Cited by 85 publications

References 42 publications

Incorporating Transformers and Attention Networks for Stock Movement Prediction

Incorporating Transformers and Attention Networks for Stock Movement Prediction

SCMFTS: Scalable and Distributed Complexity Measures and Features for Univariate and Multivariate Time Series in Big Data Environments

Fruit category classification by fractional Fourier entropy with rotation angle vector grid and stacked sparse autoencoder

Contact Info

Product

Resources

About