Comparative Study Of Classification Algorithms

After the end of WWI the roaring 20’s came to life, the economy was booming, and the U.S. felt a sense of pride. Sadly though all was not as grand as it seemed the land was stained with blood from racial issues. In the book The Learning Tree by Gordon Parks a coming to age story Newt Winger tells the tale of “how it feels to be black in the white man’s world.” He faces the struggle of the racial issue as he grows up in his little town of Cherokee Flats. The young twelve year old lives in a society where blacks are slowly being integrated but are still left with the shortest straw; not being allowed to eat or work in some places and the high schools black students are not allowed on the football nor the baseball teams while a certain teachers

I have analyzed three online news publications from different parts of the world. Those are Fox News from the U.S., BBC from England and Mail & Guardian from South Africa. In this research paper I will present quantitative research that indicates the statistically significance between them.

In this type of opinion mining, the sentiment extracted from the entire document is summarized as the opinion of the entire data. It is assumed that the document is written from the perception of a single author with a specific opinion in

Document opinion analysis is about classifying the overall document that have sentiment words expressed by the authors. The task is to determine whether a document is positive, negative or neutral (Wawre & Deshmukh). Document level categorization attempts to classify sentiments in web forum postings, blogs, movie and news

It is used to understand the emotion conveyed in a textual message. It involves identifying the opinion, extracting the features or objects for which the opinion is expressed and then categorizing the opinion as a positive, negative or neutral and thus assigning it a polarity (Liu 2010). The growth in social media provides a wider platform which has allowed for an abundance in the expression of opinions, including product reviews, blogs, and discussion groups or simply as comments and tweets. Different techniques for sentiment analysis use Natural Language processing and machine learning perform Sentiment analysis on the large quantities of data available on the social media networks.

With hundreds of millions of people sending countless hours on social media to share, communicate, connect, interact, and create user-generated data at an unprecedented rate, social media has become one unique source of big data.” The focus of the research problem under Social Media mining will be the sentimental analysis/political trends because of the sheer amount of data available on social media sites with people posting about politics.

1.3 Literature Review on Seven Data Mining Techniques . . . . . . . . . . . . . . . . . 6

There are many objectives of data mining like transform raw data into beneficial knowledge, calculation (analytical data mining is common type of data mining and has most direct business application). This makes it difficult for a potential customer to read them in order to make a decision on whether to buy the product. Usually seller who wants to sell products on the Web ask their customers to comment and opinion for the products and services. As e-commerce is becoming more and more popular, the number of customer reviews that a product receives grows rapidly. For a popular product, the number of reviews can be in hundreds. This summarization task is unlike traditional text summarization because it is only involved in the exact features of the product that customers have opinions on and also whether the opinions are positive or negative. Do not summarize the reviews by selecting or rewriting a subset of the original sentences from the reviews to capture their main points as in the classic text summarization. A number of procedures are presented to mine such features. Stage for customer’s review mining:

Naïve Bayes classifier is a simple but effective the Bayesian classifier built upon the strong assumption that different features are independent with each other. Classification is done by selecting the highest posterior of classification variable given a set of feature. Naive Bayes classifiers assume that the effect of a variable value on a given class is independent of the values of other variable. This assumption is called class conditional independence. An advantage of the naive Bayes classifier is that it requires a small amount of training data to estimate the variable values necessary for classification.

In the sentiment extraction phase, the candidate’s name extracted from the resume was used to search on Twitter to find his/her profile. The candidate’s tweets are retrieved for further processing using Tweepy package and its API [13]. Then, these tweets are analysed to extract the sentiments for each candidate. This is done by using Textblob package which is a popular natural language processer [14].

The approach used in this thesis is inspired by Bollen et al’s strategy [12], with a step taken forward to implement PageRank algorithm to increase the accuracy of results and use of different sentiment analysis techniques than the techniques used by him. In 2010, Bollen used Twitter data for finding the predictability of Twitter sentiments on stock market with high accuracy. He proposed a method for prediction of the changes in the stock market price based on the mood of people on Twitter.

In the academic journal by Fleur Laros and Jan-Benedict Steenkamp published in 2005, emotions in consumer behavior are illustrated through a hierarchy-based system of classifying. The authors suggest that the understanding and interpretation of emotion is extremely beneficial for marketers. When an individual views an advertisement, the emotions mediate the consumer 's response to the material. Depending on the emotion the consumer feels after viewing the advertisement, this can promote them to purchase that good or service or deter them away from completing the transaction (Laros & Steenkamp, 2005).

Some combined rule algorithms were proposed in (Medhat et al., 2008a). Therefore, a study on decision tree and decision rule problem is done by Quinlan (1986). Probabilistic Classifier Probabilistic classifiers make the utilization of blend of models for classification. Every class is considered to be a component of the mixed model. We have described various probabilistic classifiers for sentiment analysis problem in the next subsection. 4.1.1.4.1 Naive Bayes Classifier (NB). It is the frequently used classifier in sentiment analysis. In sentiment analysis, naive Bayes classifier calculates the posterior probability of either positive class or negative class depending on the sentiment words distributed over the document. The work of naïve Bayes classifier is based on the Bag-of-word extraction of features in which the word’s position is overlooked in the whole text. This classifier uses the Bayes theorem. It calculates the probability for the sentiment word in a document and tells whether that word belongs to the positive or negative class. The probability can be calculated using the given formula. This assumption results in Bayesian Network. A Bayesian network is a directed acyclic graph containing nodes and edges, where nodes denote the random variables and the edges denote the conditional dependencies. It is a conditional exponential classifier that takes the feature sets with label and converts them into

Support Vector Machine (SVM) is primarily a classier method that performs classification tasks by constructing hyperplanes in a multidimensional space that separates cases of different class labels. SVM supports both regression and classification tasks and can handle multiple continuous and categorical variables. For categorical variables a dummy variable is created with case values as either 0 or 1. Thus, a categorical dependent variable consisting of three levels, say (A, B, C), is represented by a set of three dummy variables:

Sentiment analysis or opinion mining is an emerging area of research, because, the impact of the web is increasing at a very fast rate, now most of the people would like to share their opinions, feelings and experiences on the web.. Now people commonly use blogs, forums, e-news, reviews channels and the social networking platforms such as