Near Real-time Prediction Research Articles

A Stacking Ensemble Deep Learning Model for Bitcoin Price Prediction Using Twitter Comments on Bitcoin

Cryptocurrencies can be considered as mathematical money. As the most famous cryptocurrency, the Bitcoin price forecasting model is one of the popular mathematical models in financial technology because of its large price fluctuations and complexity. This paper proposes a novel ensemble deep learning model to predict Bitcoin’s next 30 min prices by using price data, technical indicators and sentiment indexes, which integrates two kinds of neural networks, long short-term memory (LSTM) and gate recurrent unit (GRU), with stacking ensemble technique to improve the accuracy of decision. Because of the real-time updates of comments on social media, this paper uses social media texts instead of news websites as the source data of public opinion. It is processed by linguistic statistical method to form the sentiment indexes. Meanwhile, as a financial market forecasting model, the model selects the technical indicators as input as well. Real data from September 2017 to January 2021 is used to train and evaluate the model. The experimental results show that the near-real time prediction has a better performance, with a mean absolute error (MAE) 88.74% better than the daily prediction. The purpose of this work is to explain our solution and show that the ensemble method has better performance and can better help investors in making the right investment decision than other traditional models.

Optimal Estimation MSG-SEVIRI Clear-Sky Total Column Water Vapour Retrieval Using the Split Window Difference

A new algorithm for the retrieval of day-time total column water vapour (TCWV) from measurements of a MSG-SEVIRI (Meteosat Second Generation Spinning Enhanced Visible and Infrared Imager) instrument is presented. The retrieval is based on a forward operator, at the core of which lies Radiative Transfer for TIROS Operational Vertical Sounder (RTTOV). This forward model relates TCWV and surface temperature to brightness temperatures in the split window at 11 and 12µm with the use of a first guess for temperature and humidity profiles from the ERA5 reanalysis. The forward model is then embedded in a full Optimal Estimation (OE) method, which yields pixel by pixel uncertainty estimates and performance indicators. The algorithm is applicable to any instrument which features the split window configuration, given a first guess for atmospheric conditions (i.e., from NWP) and an estimate of surface emissivity at 11 µm. The algorithm was developed within the framework of RealPEP (Near-Realtime Quantitative Precipitation Estimation and Prediction) in which the advancement of the estimation and nowcasting of extreme precipitation and flooding in Germany are studied. Thus, processing and validation has been limited to the German domain. Three independent ground-based TCWV observation data sets were used as reference, i.e., AERONET (Aerosol Robotic Network), GNSS Germany (Global Navigation Satellite System) and measurements from two MWR (Microwave Radiometer) sites. The validation concludes with good agreement, with absolute biases between 0.11 and 2.85 kg/m2, root mean square deviations (rmsds) between 1.63 and 3.24 kg/m2 and Pearson correlation coefficients ranging from 0.96 to 0.98. The retrievals uncertainty estimates were evaluated against AERONET. The comparison suggests that, in sum, uncertainties are estimated well, while still some error sources seem to be over- and underestimated. In limited case studies it could be shown that SEVIRI TCWV is capable to both display large scale variabilities in water vapour fields and reproduce the daily course of water vapour exposed by ground-based observations.

Investigation of the response time of the equatorial ionosphere in context of the equatorial electrojet and equatorial ionization anomaly

Abstract. Equatorial Electrojet (EEJ) and Equatorial Ionization Anomaly (EIA) are two large-scale processes in the equatorial/low latitude ionosphere, driven primarily by the eastward electric field during daytime. In the present paper we investigate the correlation between the Integrated EEJ strength (IEEJ) and the EIA parameters like the total electron content at the northern crest, location of crest in Magnetic latitude and strength of the EIA for the Indian sector. A good correlation has been observed between the IEEJ and EIA when a time delay is introduced between IEEJ and EIA parameters. This time delay is regarded as the response time of equatorial ionosphere in context of the evolution of EIA vis-à-vis EEJ. Further, a seasonal variation in the time delay has been observed, which is believed to be due to changes in thermospheric wind. Using the response time and the linear relationship obtained, the possibility of near-real time prediction of EIA parameters has been attempted and found that the prediction holds well during the geomagnetically quiet periods. The paper discusses these aspects in detail.

Analytical representation of izmem model for near-real time prediction of electromagnetic weather

Abstract In this study we describe an analytical representation of the IZMIRAN Electrodynamic Model (IZMEM) that allows easy calculation of a number of ionospheric parameters for any point above the corrected geomagnetic latitude φ ⩾ 57° over the Northern and Southern polar regions and for all seasons of a year. These are the distributions of ionospheric electric potentials, electric fields, field-aligned and ionospheric currents, horizontal ion drift velocities, Joule heating rate, and ground magnetic disturbances. Comparisons of the modeling results with the Heppner–Maynard and Foster ionospheric convection models as well as with various satellite observations are also presented and discussed. A simple, inexpensive scheme is suggested for specifications and forecast of polar ionospheric convection patterns solely from interplanetary observations made at the Lagrangian L1 point upstream in the solar wind. © 1998 Elsevier Science Ltd. All rights reserved.