Machine Learning-based USD/PKR Exchange Rate Forecasting Using Sentiment Analysis of Twitter Data

This study proposes an approach based on machine learning to forecast currency exchange rates by applying sentiment analysis to messages on Twitter (called tweets). A dataset of the exchange rates between the United States Dollar (USD) and the Pakistani Rupee (PKR) was formed by collecting information from a forex website as well as a collection of tweets from the business community in Pakistan containing finance-related words. The dataset was collected in raw form, and was subjected to natural language processing by way of data preprocessing. Response variable labeling was then applied to the standardized dataset, where the response variables were divided into two classes: “1” indicated an increase in the exchange rate and “ −1” indicated a decrease in it. To better represent the dataset, we used linear discriminant analysis and principal component analysis to visualize the data in three-dimensional vector space. Clusters that were obtained using a sampling approach were then used for data optimization. Five machine learning classifiers—the simple logistic classifier, the random forest, bagging, naïve Bayes, and the support vector machine—were applied to the optimized dataset. The results show that the simple logistic classifier yielded the highest accuracy of 82.14% for the USD and the PKR exchange rates forecasting.     

Economic impacts and challenges of China’s petroleum industry: An input–output analysis

It is generally acknowledged that the petroleum industry plays an important role in China’s national economic and social development. The direct, indirect, and induced impacts of China’s petroleum industry are analyzed in this study by using the Input-Output approach. The study also considers the main challenges that China’s economy might face in the future. The research results suggest the following: (1) The total economic impacts coefficients on output, given each unit of final demands change in extraction of petroleum and processing of petroleum, are 1.9180 and 3.2747 respectively, and the corresponding economic impacts coefficients on GDP are 1.0872 and 0.9001 respectively; (2) Extraction of petroleum has a more direct impact on GDP, while processing of petroleum has a greater effect on the total output; (3) Extraction of petroleum’s total economic impacts coefficients on both output and GDP have remained stable in recent years after a period of long decline; processing of petroleum’s total economic impacts coefficient on output is steadily increasing; (4) Import uncertainty, the likelihood of rising oil prices, and net oil exports caused by items manufactured with petroleum products (i.e. “Made in China” goods) are the main challenges the petroleum industry will cause for China’s overall economy.     

Mechanical properties of amorphous silicon carbonitride thin films at elevated temperatures

The mechanical properties of amorphous silicon carbonitride (a-SiC _x N _y ) films with various nitrogen content (y = 0–40 at.%) were investigated in situ at elevated temperatures up to 650 °C in inert atmosphere. A SiC film was measured also at 700 °C in air. The hardness and elastic modulus were evaluated using instrumented nanoindentation with thermally stable cubic boron nitride Berkovich indenter. Both the sample and the indenter were separately heated during the experiments to temperatures of 300, 500, and 650 °C. Short duration high temperature creep tests (1200 s) of the films were also carried out. The results revealed that the room temperature hardness and elastic modulus deteriorate with the increase of the nitrogen content. Furthermore, the hardness of both the a-SiC and the a-SiCN films with lower nitrogen content at 300 °C drops to approx. 77 % of the corresponding room temperature values, while it reduces to 69 % for the a-SiCN film with 40 at.% of nitrogen. Further increase of temperature is accompanied with minor reduction in hardness except for the a-SiCN film with highest nitrogen content, where the hardness decreases at a much faster rate. Upon heating up to 500 °C, the elastic modulus of the a-SiCN film decreases, while it increases at 650 °C due to the pronounced effect of short-range ordering. The steady-state creep rate increases at elevated temperatures and the a-SiC exhibits slower creep rates compared to the a-SiCN films. The value of the universal constant x = 7 relating the W _p/W _t and H/E ^* was established and its applicability was demonstrated. Analysis of the experimental indentation data suggests a theoretical limit of hardness to elastic modulus ratio of 0.143.     

Packet Recycling and Delayed ACK for Improving the Performance of TCP over MANETs

Most of the schemes that were proposed to improve the performance of transmission control protocol (TCP) over mobile ad hoc networks (MANETs) are based on a feedback from the network, which can be expensive (require extra bandwidth) and unreliable. Moreover, most of these schemes consider only one cause of packet loss. They also resume operation based on the same stand-by parameters that might vary in the new route. Therefore, we propose two techniques for improving the performance of TCP over MANETs. The first one, called TCP with packet recycling (TCP-PR), allows the nodes to recycle the packets instead of dropping them after reaching the retransmission limit at the MAC layer. In the second technique, which is called TCP with adaptive delay window (TCP-ADW), the receiver delays sending TCP ACK for a certain time that is dynamically changed according to the congestion window and the trip time of the received packet. TCP-PR and TCP-ADW are simple, easy to implement, do not require network feedback, compatible with the standard TCP, and do not require distinguishing between the causes of packet loss. Our thorough simulations show that the integration of our two techniques improves the performance of TCP over MANETs.     

Symbolic analysis of large-scale networks using a hierarchical signal flowgraph approach

This paper presents an approach suitable for the symbolic analysis of large-scale active networks. The method depends on the partitioning of the network into smaller networks which are then symbolically analyzed noniteratively using the Mason's signal flowgraph models of each partition. The resulting solutions, which are reduced signal flowgraphs (transfer functions) for the subnetworks are then hierarchically combined to produce the final solution or solutions (transfer functions) for the entire system. The advantage of such an approach is the reduction in the number of symbolic terms compared to the conventional approaches, and the ability to analyze hybrid systems consisting of electrical and nonelectrical parts. The result of the analysis is a series of equations that have an upward hierarchical dependency on each other.     

Adaptive cross-layer MAC design for improved energy-efficiency in multi-channel wireless sensor networks

We present a novel cross-layer design for improving energy efficiency in a wireless sensor network that utilizes a multi-channel non-persistent CSMA MAC protocol with adaptive MQAM modulation at the physical layer. Cross-layer interactions are achieved through joint, traffic-dependent adaptation of the backoff probability at the MAC layer and the modulation order at the physical layer. The joint optimization of the backoff probability and the modulation order is conducted subject to a constraint on the packet retransmission delay. Such an optimization is shown to produce a significant improvement in the per-bit energy requirement for successful packet delivery. Our analytical findings are verified through numerical results and computer simulations.     

Effect of the preparation method on the surface coverage of Mo/Al2O3 catalysts

Two series of Mo/Al₂O₃ catalysts were prepared by equilibrium adsorption and incipient wetness impregnation methods. The effect of preparation method on the surface coverages of the calcined catalysts was investigated by the combined use of CO₂ chemisorption, low temperature CO adsorption and ion scattering spectroscopy (ISS). For a given Mo loading, the CO₂ and CO adsorption results showed little difference between the two preparation methods. As previously noted, the CO₂ chemisorption method overestimated the Mo surface coverage. In contrast to the adsorption methods, the ISS technique gave different Mo surface coverage values for a given Mo content of the two series of catalysts. This apparent discrepancy was attributed to different repartition of the Mo phase between the internal and external surfaces which can only be detected by ISS. This interpretation is supported by the observed agreement between the coverage values measured from ISS and low temperature CO adsorption for presumably uniform catalysts obtained by the equilibrium adsorption method.     

Error-resilient packet-switched mobile video telephony with channel-adaptive rateless coding and early reference picture selection

Providing high-quality video for packet-switched wireless video telephony on handheld devices is a challenging task due to packet loss, varying bandwidth, and end-to-end delay constraints. While many error resilience techniques have been proposed for video transmission over wireless channels, only a few were specifically designed for mobile video telephony. We propose a low-complexity channel-adaptive error resilience technique for packet-switched mobile video telephony, which combines rateless coding, feedback, and reference picture selection. In contrast to previous approaches, our technique uses cumulative feedback at every transmission opportunity and predicts when decoding is likely to fail so that reference picture selection can be triggered at an early stage. Experimental results for H.264 video sequences show that the proposed technique can achieve improvements of 1.64 dB in peak signal-to-noise ratio over benchmark techniques in simulated Long-Term Evolution networks.     

Modeling the Effects of Coolant Application in Friction Stir Processing on Material Microstructure Using 3D CFD Analysis

The ability to generate nano-sized grains is one of the advantages of friction stir processing (FSP). However, the high temperatures generated during the stirring process within the processing zone stimulate the grains to grow after recrystallization. Therefore, maintaining the small grains becomes a critical issue when using FSP. In the present reports, coolants are applied to the fixture and/or processed material in order to reduce the temperature and hence, grain growth. Most of the reported data in the literature concerning cooling techniques are experimental. We have seen no reports that attempt to predict these quantities when using coolants while the material is undergoing FSP. Therefore, there is need to develop a model that predicts the resulting grain size when using coolants, which is an important step toward designing the material microstructure. In this study, two three-dimensional computational fluid dynamics (CFD) models are reported which simulate FSP with and without coolant application while using the STAR CCM+ CFD commercial software. In the model with the coolant application, the fixture (backing plate) is modeled while is not in the other model. User-defined subroutines were incorporated in the software and implemented to investigate the effects of changing process parameters on temperature, strain rate and material velocity fields in, and around, the processed nugget. In addition, a correlation between these parameters and the Zener-Holloman parameter used in material science was developed to predict the grain size distribution. Different stirring conditions were incorporated in this study to investigate their effects on material flow and microstructural modification. A comparison of the results obtained by using each of the models on the processed microstructure is also presented for the case of Mg AZ31B-O alloy. The predicted results are also compared with the available experimental data and generally show good agreement.