Priyank Thakkar

Predicting stock and stock price index movement using Trend Deterministic Data Preparation and machine learning techniques

Four machine learning algorithms are used for prediction in stock markets.Focus is on data pre-processing to improve the prediction accuracy.Technical indicators are discretised by exploiting the inherent opinion.Prediction accuracy of algorithms increases when discrete data is used. This paper addresses problem of predicting direction of movement of stock and stock price index for Indian stock markets. The study compares four prediction models, Artificial Neural Network (ANN), Support Vector Machine (SVM), random forest and naive-Bayes with two approaches for input to these models. The first approach for input data involves computation of ten technical parameters using stock trading data (open, high, low & close prices) while the second approach focuses on representing these technical parameters as trend deterministic data. Accuracy of each of the prediction models for each of the two input approaches is evaluated. Evaluation is carried out on 10years of historical data from 2003 to 2012 of two stocks namely Reliance Industries and Infosys Ltd. and two stock price indices CNX Nifty and S&P Bombay Stock Exchange (BSE) Sensex. The experimental results suggest that for the first approach of input data where ten technical parameters are represented as continuous values, random forest outperforms other three prediction models on overall performance. Experimental results also show that the performance of all the prediction models improve when these technical parameters are represented as trend deterministic data.

Predicting stock market index using fusion of machine learning techniques

Two stage fusion model comprising three machine learning techniques is used.Emphasis is on adequacy of information given to prediction models.First stage provides future value of statistical parameters helping the later stage.Two stage fusion model helps in decreasing overall prediction error. The paper focuses on the task of predicting future values of stock market index. Two indices namely CNX Nifty and S&P Bombay Stock Exchange (BSE) Sensex from Indian stock markets are selected for experimental evaluation. Experiments are based on 10years of historical data of these two indices. The predictions are made for 1-10, 15 and 30days in advance. The paper proposes two stage fusion approach involving Support Vector Regression (SVR) in the first stage. The second stage of the fusion approach uses Artificial Neural Network (ANN), Random Forest (RF) and SVR resulting into SVR-ANN, SVR-RF and SVR-SVR fusion prediction models. The prediction performance of these hybrid models is compared with the single stage scenarios where ANN, RF and SVR are used single-handedly. Ten technical indicators are selected as the inputs to each of the prediction models.

Opinion Spam Detection Using Feature Selection

In modern times, it has become very essential for e-commerce businesses to empower their end customers to write reviews about the services that they have utilized. Such reviews provide vital sources of information on these products or services. This information is utilized by the future potential customers before deciding on purchase of new products or services. These opinions or reviews are also exploited by marketers to find out the drawbacks of their own products or services and alternatively to find the vital information related to their competitors products or services. This in turn allows to identify weaknesses or strengths of products. Unfortunately, this significant usefulness of opinions has also raised the problem for spam, which contains forged positive or spiteful negative opinions. This paper focuses on the detection of deceptive opinion spam. A recently proposed opinion spam detection method which is based on n-gram techniques is extended by means of feature selection and different representation of the opinions. The problem is modelled as the classification problem and Naïve Bayes (NB) classifier and Least Squares Support Vector Machine (LS-SVM) are used on three different representations (Boolean, bag-of-words and term frequency -- inverse document frequency (TF-IDF) ) of the opinions. All the experiments are carried out on widely used gold-standard dataset.

Investigating the Effect of Varying Window Sizes in Speaker Diarization for Meetings Domain

Speaker Diarization deals with determining “who spoke when?” with the help of computers. It is extremely useful in speech transcription, subtitle generation and extracting opinions among others. Diarization is a process in which a relatively long audio recording is processed and speech segments are labeled using respective speaker identities. Such systems are also helpful in determining number of speakers in any conversation or meeting. This field of research is active since long and researchers have been successful to improve the system over time. In this paper, we have made an attempt to study the effect of varying window sizes and threshold criteria on performance of speaker diarization system. Experiments are conducted using LIUM (Laboratoire d’Informatique de l’Universite du Maine) toolkit on Augmented Multi-party Interaction (AMI) meeting data corpus. The proposed approach has shown promising results in case of significantly less number of frames per window.

Comparative analysis of zoning based methods for Gujarati handwritten numeral recognition

Gujarati is one of the ancient Indian languages spoken widely by the people of Gujarat state. This paper is concerned with the recognition of handwritten Gujarati numerals. For recognition of Gujarati numerals zoning based Feature extraction method is used. Numeral image is divided in 16×16, 8×8, 4×4 and 2×2 Zones. After feature extraction through the zoning method, Naive Bayes classifier and multilayer feed forward neural network classifier are implemented for the classification of numerals. For the database generation, 14,000 samples of each numeral are used. The overall recognition rates of this method used for recognition of Gujarati numeral using 16×16, 8×8, 4×4 and 2×2 zoning with neural network are 93.03%, 95.92%, 91.89% and 61.78% and with Naive Bayes classifier are 75%, 85.60%, 81% and 53.75% respectively.

Routing Protocols for Wireless Multimedia Sensor Networks: Challenges and Research Issues

Due to miniaturization of hardware and availability of low-cost, low-power sensors, Wireless Sensor Network and Multimedia Sensor Network applications are increasing day by day. Each application has a specific quality of service and experience requirements. The design of routing and MAC protocol which can fulfill the requirements of the application is challenging given the constrained nature of these devices. Considerable efforts are directed towards the design of energy efficient QoS-aware routing protocols. In this article, we present state of the art review of routing protocols for Wireless Multimedia Sensor Networks while addressing the challenges and providing insight into research issues.

Outcome Fusion-Based Approaches for User-Based and Item-Based Collaborative Filtering

Collaborative Filtering (CF) is one of the most effective approaches to engineer recommendation systems. It recommends those items to user which other users with related preferences and tastes liked in the past. User-based and Item-based Collaborative Filtering (IbCF) are two flavours of collaborative filtering. Both of these methods are used to estimate target user’s rating for the target item. In this paper, these methods are implemented and their performance is evaluated on the large dataset. The major attention of this paper is on exploring different ways in which predictions from UbCF and IbCF can be combined to minimize overall prediction error. Predictions from UbCF and IbCF are combined through simple and weighted averaging and performance of these fusion approaches is compared with the performance of UbCf & IbCF when implemented individually. Results are encouraging and demonstrate usefulness of fusion approaches.

Gujarati handwritten numeral recognition through fusion of features and machine learning techniques

Languages have played a major role in Indian history and they continue to influence the lives of Indians till date. Plentiful research on optical character recognition (OCR) techniques for Indian languages such as Hindi, Tamil, Bangla, Kannada, Gurumukhi, Malayalam and Marathi has already been carried out. Research efforts on Gujarati character recognition are few and yet to gain momentum. This paper intends to bring Gujarati character recognition in attention. Methods based on artificial neural network (ANN), support vector machine (SVM) and naive Bayes (NB) classifier are exercised for handwritten Gujarati numerals recognition. Experiments are carried out on two large datasets using three different kinds of features and their fusion. Zone-based, projection profiles-based and chain code-based features are employed as individual features. The paper proposes to use a fusion of these features for learning prediction models. Experimental results show significant improvement over state-of-the-art and validate our proposals.

Surveying stock market portfolio optimization techniques

Optimizing a stock market portfolio requires decision making at two distinct stages, first is to select the stocks and second is to assign distribution of investment amount among these selected stocks. Given the historical data of stocks, the role of optimization models is to select stocks and assign portfolio proportion to the selected stocks. Selection and weight assignment to stocks are co-occurring activities. Investors prime motive is to maximize the return and minimize the risk of portfolio. Stock market is uncertain and volatile and therefore, Artificial Intelligence, Machine Learning and Soft Computing techniques are viable candidates which can help in optimization and making decisions using such data. This paper surveys the research carried out in the domain of stock market portfolio optimization. Paper compares research efforts in the domain on the basis of techniques used, risk models and stock markets considered. It is observed from the surveyed papers that Artificial Intelligence, Machine Learning and Soft Computing techniques are widely accepted for studying and evaluating stock market behavior and optimizing portfolios.