Sentiment Analysis or Opinion Mining for Stock Market Prediction.


•Sentiment analysis or opinion mining make use of text mining,natural languaging processing(NLP), in order to identify and extract the subjective content by analyzing user’s opinon, evaluation, attitudes, sentiments and emotions.

•Sentiment analysis for stock market indicators such as Sensex and Nifty has been done to predict the stock price.

•It is very important for investors to predict the Stock Market before investing in it. In which Sentiment analysis helps a lot the way we do business.

•Experimental results have verified that proposed algorithm can provide various numbers of conclusion and provide suggestion for future work .

Sentiment analysis is a technique used to extract the intelligent information based on the person’s opinon raw data available in the internet by classifying them positive,negative and natural sentiment and consider these feedback for improvement of policies,planning and process for product development.

The Basic Approach is as Follows :

•A Word expressed as according to user Opinon can be Positive or Negative.

•Indian Stock Market gained the Interest of Investors investing in two main Stock Market such as BSE(Bombay Stock Exchange) and NSE(National Stock Exchange) respectively.

•To Predict the Indian Stock Market we Use these Indicator’s i.e Sensex(BSE 30 Companies) and Nifty(NSE 50 Companies).

Sentiment analysis technique can be classified into Machine Learning Approach and Lexicon Based Approach.These Techniques are Explained as follows:-

•1.Machine Learning Approach

Machine learning techniques that can be applied in the field of sentiment analysis can be divided as Supervised and Unsupervised Learning.

•1.1 Unsupervised Learning

This has no explicit target output associated with input,and it is learning through observation.Famous approach is Clustering,in which similarities of elements in the training data is found out.Cluster similarity parameter is defined upon metrics such as HMV,K-means,Euclidean distance,Gaussian mixture models,Self- organizing maps and Hierarchical are some of the Clustering algorithms.

1.2 Supervised Learning

This basically make use of known datasets to make the prediction for output result classified and divided into training set and test set. Training set is used for the for learning different properties of documents and Test set is used for evaluating the performance classifier.Various Supervised learning techniques are used as follows.

1.2.1 Decision tree classifier

Decision tree uses a hierarchical decomposition of training data in which data is divided based on the condition of attribute values.Generally a condition is presence or absence of words.Each non-leaf node is associated with attribute and leaf node is associated with the classification value positive or negative.

1.2.2 Rule based classifier

This classfier consider the emotions of text and consider them positive or negative based on them.This rule is same as fuzzy-systems which allow the intermediate value to be defined between conventional evaluation like yes/no,high/low,true/false,etc.

1.2.3 Probabilistic classifier

Probabilistic classifier are developed by assuming generative models which are product distributions over the original attribute space or more involved spaces.Probabilistic classifier is based on the prediction of input given probability distribution.Two most important probabilistic classifier are as follows. Naive Bayes classifier

This is based on Bayes theorem of probabilistic model.In this we tried to estimate the probability of a text whether it is positive/negative class. Maximum Entropy classifier

This is probabilistic based classifier which belong to the exponential model class..Principle of maximum entropy is used in this chapter and distribution having largest entropy is chosen

1.2.4 Linear classifier

This is one that partition the set into it’s respective domain with the help of a line and it partitioning with the curve is to be called as hyper plane.Two more Linear classifier as follows : Support Vector Machine(SVM)

It is widely used for classification and regression analysis.The basic idea is to determine linear seperator in search space which can separate the different classes. Neural Network

This offers nonlinearity,input output mapping,adaptivity and fault tolerance.

1.2.5 Lexicon based approach

This method is based on that performs sum of the seniment orientation of each word makes contextual sentiment orientation.This further divide into two types as follows: Dictionary based approach

Felling of happy,sad or depressed can be found out by comparing word against lexicons from dictionaries. Corpus based approach

This approach is based on seeding list of opinion then comparing with other features of opinion which have similar context.

1.2.6 Hybrid Fuzzy Neural Network based learning

This is one which is based on hybrid system technology to solve analysis problem for indain stock market prediction.classification are 1)Sequential hybrids:pipelike structure. 2)auxiliary hybrids:subroutine manipulation .3)Embedded hybrids:auxiliary but intertwined.

•For Implementation purpose Sensex and Nifty data has been fetched out from live server data i.e timesofindia.

•Python script have been choosen to run live sensex and nifty data in time count interval of 1 second and for that sleep time have been choosen.

•After implementaion result should be drawn In different time interval data has been fetched out in which Sensex and Nifty remains constant.

•Indian Stock Market is Unstable before investing we should regularly test it in which sentiment analysis plays an important role.

•This method provides fetching of live data of Sensex and Nifty which helps in prediction of indian stock market which is done with the help of python scripting language in various time interval.

Future Work

In future we can use more advanced functions of python script code to do Sentiment Analysis for Indian Stock Market Prediction .