Analysis of Stock Market Trend using Integrated Clustering and Weighted Rule Mining Technique

Analysis of Stock Market Trend using Integrated Clustering and Weighted Rule Mining Technique S.Karthik #, K.K.Sureshkumar * # M.Phil - Computer Science (Part Time), Research Scholar, Kongu Arts and Science College, Erode, Tamil Nadu, India. * Assistant Professor, P.G.Department of Computer Science, Kongu Arts and Science College, Erode, Tamil Nadu, India. Abstract - Stock markets manages huge amount of data values. Thousands of companies are traded through the stock markets. Market trade flows are reflected using the market indexes and sector indexes. BSE SENSEX and NIFTY are two major indexes in Indian stock exchanges. For predicting and analyzing the market trends data mining techniques are also used. Statistical techniques are used for the market price prediction process. Inaccurate results are produced in the statistical analysis. To find the stock market trend with index dependency analysis environment hybrid clustering and association algorithm is not appropriate. Statistic analysis techniques are not suitable for trend analysis with index relationship. The stock market transactions data is analyzed with clustering and weighted rule mining techniques. The K-means clustering algorithm is used to cluster the transaction with respect to the market flows. The market trade transactions are divided into three zones such as up trend, down trend and stable zone. The weighted rule mining technique is applied to fetch patterns from the indexes, sector indexes and company price values. Apriori algorithm is modified to carry out weighted rule mining process. The system produces the market trade trend flow with market indexes and sector index values. Keywords - Data Mining, Apriori Algorithm, Integrated Clustering, Weighted Rule Mining, Stock Market, National Stock Exchange (NSE), Bombay Stock Exchange (BSE), and Trend Analysis. I. INTRODUCTION Nowadays data mining is primarily used by companies with a strong consumer focus on retail, financial, communication and marketing organizations. It enables these companies to determine relationship among internal factors such as price, product positioning or staff skills and external factors such as economic indicators, competition and customer demographics. It enables them to determine the impacts on sales, customer satisfaction and corporate profits. Finally, it enables them to drill down into summary information to view detailed transactional data. Stock exchange means a market place for purchase and sale of industrial and financial securities. The word stock means a fraction of the company s capital and the word exchange means a place for purchasing and selling something. In actual, the term is used in a wider sense [13]. There are leading Stock exchanges in India are as Bombay Stock Exchange and National Stock Exchange (NSE). Bombay Stock Exchange is the oldest stock exchange in Asia with a rich heritage, now spanning three centuries in its 135 years of existence. Now popularly known as BSE was established as "The Native Share & Stock Brokers' Association" in 1875. BSE is the first stock exchange in the country which obtained permanent recognition from the Government of India under the Securities Contracts Act 1956. BSE's pivotal and preeminent role in the development of the Indian capital market is widely recognized [5]. It deals with the shares listed in the exchange. Here all types of shares are bought and sold in the market. NSE was established in 1994. This exchange deals with all types of shares in old and new economic shares [6]. This exchange trading point is called NIFTY. In the recent days the stock exchange campus is the centralized point for the share market. The remainder of the paper is organized as follows. Section II reviews the literature in prediction of the stock trend by data mining. Section III focuses on the objectives of the research. Section IV discusses about the data mining in stock market analysis. Section V discusses about the methodology and implementation of stock trend analysis. The findings and outcome has reveled in Section VI. Finally, conclusions and future work were presented in Section VII. II. LITERATURE REVIEW The stock market is a complex, no stationary, chaotic and non-linear dynamic system. Forecasting stock market, currency exchange rate, bank bankruptcies, understanding and managing financial risk, trading futures, credit rating, loan management, bank customer profiling, and money laundering analyses are core financial tasks for data mining [10]. Decision trees are excellent tools for making financial or number based decisions where a lot of complex information needs to be taken into account. They provide an effective structure in which alternative decisions and the implications of taking those decisions can be laid down and evaluated. They also help to form an accurate, balanced picture of the risks and rewards that can result from a particular choice. In a stock market, how to find right stocks and right timing to buy has been of great interest to investors. To achieve this objective, Muh-Cherng et al. present a stock trading method S.Karthik et.al. 984 www.ijcsmr.org

by combining the filter rule and the decision tree technique. The filter rule, having been widely used by investors, is used to generate candidate trading points. These points are subsequently clustered and screened by the application of a decision tree algorithm. Compared to previous literature that applied such a combination technique, this research is distinct in incorporating the future information into the criteria for clustering the trading points [11]. Taiwan and NASDAQ stock markets are used to justify the proposed method. Experimental results show that the proposed trading method outperforms both the filter rule and the previous method. The appropriate stock selections those are suitable for investment is a difficult task. The key factor for each investor is to earn maximum profits on their investments. Numerous techniques used to predict stocks in which fundamental and technical analysis are one among them. Artificial Neural networks are potentially useful for studying the complex relationships between the input and output variables in the system. The stock exchange operations can greatly benefit from efficient forecast techniques. Some prediction algorithms and functions are used to predict future share prices and their performance will be compared. In this, Sureshkumar and Elango have applied different neural classifier functions by using the Weka tool. Different prediction approaches have been applied such as Gaussian processes, isotonic regression, least mean square, linear regression, multilayer perceptron, pace regression, simple linear regression and SMO regression. In this eight functions taken into consideration for analysis to predict values and evaluated. The results from analysis shows that isotonic regression function offers the ability to predict the stock prices more accurately than the other existing techniques [15]. By comparing the results of the correlation coefficient values and error percentage the isotonic regression is the best suited method for predicting the stock prices. Listed companies financial distress prediction is important to both listed companies and investors. Jie and Hui present a data mining method combining attribute-oriented induction, information gain, and decision tree, which is suitable for preprocessing financial data and constructing decision tree model for financial distress prediction. On the basis of financial ratios attributes and one class attribute, adopting entropy-based discrimination method, a data mining model for listed company s financial distress prediction is designed [8]. The empirical experiment with 35 financial ratios and 135 pairs of listed companies as initial samples got satisfying result, which testifies to the feasibility and validity of the proposed data mining method for listed companies financial distress prediction. Accurately, forecasting stock prices has been extensively studied. Jar-Long and Shu-Hui Chan provided a proposal to use a two-layer bias decision tree with technical indicators to create a decision rule that makes buy or not buy recommendations in the stock market. A novel method designed for using two-layer bias decision tree to improve purchasing accuracy [7]. Comparison with random purchases, the results indicate the system presented here not only has excellent out-of sample forecasting performance, but also delivers a significant improvement in investment returns for all listed companies. Additionally, the proposed system has few parameter requirements, stable learning, and fast learning speed. Increasingly, the system presented here has high accuracy given large amounts of varied test data, with testing periods that experienced structural change including both bull and bear markets. Based on all of the above, they believe the proposed bias decision model is very flexible, modular and easily understandable. Chi-Lin Lu and Ta-Cheng Chen have employed decision tree-based mining techniques to explore the classification rules of information transparency levels of the listed firms in Taiwan s stock market. The main purpose of their study is to explore the hidden knowledge of information disclosure status among the listed companies in Taiwan s stock market [3]. The classification accuracy has been improved by using multileaner model. In particular, the extracted rules from the data mining approach can be developed as a computer model for the prediction or classification of good/poor information disclosure potential. By using the decision tree-based rule mining approach, the significant factors with the corresponding equality/inequality and threshold values were decided simultaneously, so as to generate the decision rules. Unlike many mining approaches applying neural networks related approaches in the literature, the decision tree approach is able to provide the explicit classification rules [4]. Moreover, a multi-learner model constructed by boosting ensemble approach with decision tree algorithm has been used to enhance the accuracy rate in this work. Based on the extracted rules, a prediction model has then been built to discriminate good information disclosure data from the poor information disclosure data with great precision. Moreover, the results of the experiment have shown that the classification model obtained by the multi-learner method has higher accuracy than those by a single decision tree model. It indicates that the multi-learner model is appropriate to elicit and represent experts decision rules, and thus it has provided effective decision supports for judging the information disclosure problems in Taiwan s stock market. By using the rule based decision models, investors and the public can accurately evaluate the corporate governance status in time to earn more profits from their investment. It has a great meaning to the investors, because only prompt information can help investors in correct investment decisions. III. OBJECTIVES OF THE RESEARCH The main objective of this research is to analyze the stock market trend using integrated clustering and weighted rule mining technique. Data is very important for every organization and business. Data that was measured in gigabytes until recently, is now being measured in terabytes, and will soon approach the beta byte range. In order to achieve the goals, there is a need to fully exploit this data by extracting all the useful information from it. Unfortunately, the size and complexity of the data is such that it is impractical to manually analyze, explore, and understand the S.Karthik et.al. 985 www.ijcsmr.org

data. As a result, useful information is often overlooked, and the potential benefits of increased computational and data gathering capabilities are only partially realized. Data mining techniques like clustering and associations can be used to find meaningful patterns for future predictions. Clustering is used to generate groups of related patterns, while association provides a way to get generalized rules of dependent variables. Patterns from a huge stock data on the basis of these rules can be obtained. IV. DATA MINING IN STOCK MARKET ANALYSIS A. Merits of Integrated Clustering and Weighted Rule Minding Model The share market trend analysis system is designed to analyze the real market data for a period of 3 years from 2005 to 2007. All the data values are collected from the web site. The system produces the clustering process and trend relations analysis is done with the relationship between the index and company. Market index, sector index and company price ratio values are used in the trade prediction process. The following are the advantages of the proposed system. Trade prediction produces the price boundary for the company. Accurate price and trend prediction mechanism. The system supports both National Stock Exchange (NSE) and Bombay Stock Exchange (BSE) information. Index dependency of company is analyzed by the system. The system database can be updated with any number of data values. Efficient trade zone separation and relations analysis. B. Clustering and Weighted Rule Mining Model for Stock Market Analysis The proposed system is designed to perform the trend analysis on the market index data, sector index data and the company data values. Market data collection is the first step in the system. The sector index refers the index for the sectors such as information technology, banking and automobile. The data values are imported to the access database for the analysis. Market index data, sector index data and company price data values are optimized to group the data values. The price values are collected in five categories. Previous closing price, open price, high price, low price and quantity data values are used for the analysis. The changes in the market index; sector index and company data values are used for the trade relationship analysis. Trade prediction is carried out under the trend analysis phase [2]. Boundary prices and trend details are produced as results. The trend prediction system for stock markets is designed to analyze the Indian stock markets. The system is divided into four major modules. They are market data, clustering process, trend relations and trade prediction process. The market data module maintains the market data, sector index data and company price information. The clustering process is applied on the preprocessed data values. Trade relationship is extracted under the trade relation s module. The trend prediction module predicts the trade trend pattern with respect to the company name and trade data values. The changes in market index, sector index and company price change are analyzed with their relationships. The clustering techniques are used to group up the related trade transactions. The similar data values are arranged in separate partitions. The traded price differences are denoted as weight values [12]. The association mining is carried out on the partitions with the weight values. The weighted rule mining is performed with trade frequencies and weight values. The integrated clustering and weighted rule mining technique is used for the data partitioning and rule mining process. C. Integrated Clustering and Association Mining The data clustering techniques are used to group up the data values. The patterns are identified using the rule mining techniques. The pattern extraction process can be applied on data partitioned under the clustering process. In this method the patterns are extracted with more accurately with weight information. The attribute relationships are considered in the clustering process. This method reduces the computational overhead and time for the pattern extraction process. The stock market data values are partitioned with respect to the transaction level details. The connection with the index and sector index are also considered in the pattern identification process. Data cleaning, clustering and rule mining operations are performed under the system [9]. The clustering process produces the data partitions as the output. The data partitions are used as the input for the rule mining process. The rule mining process identifies the trends in the market data values. D. Clustering and Frequency Estimation The proposed approach is a two phased model. First the system generates clusters using K-Mean algorithm, and then MFP is designed for counting frequencies of items under their specified attributes. The sample data is collected from real share market. The system processes the data to remove the noise first, so the incomplete, missing and irrelevant data are removed and formatted according to the required format. E. K-MEANS K-means is a typical clustering algorithm and has used for classification of data for decades. Proximity is usually measured by some sort of distance; the most commonly being used is the Euclidean distance. 1 2 dist ( i, j) = ( x ik x jk ) (1) k 1 The main idea is to define k centroids, one for each cluster. These centroids should be placed in a cunning way because of different location causes different result. So, the better choice is to place them as much as possible far away from each other. This algorithm aims at minimizing an objective function, in this case a squared error function. S.Karthik et.al. 986 www.ijcsmr.org

J Where k k = j = 1 i= 1 ( j) i x c j 2. (2) ( j) 2 xi c is a chosen distance measure j The objective function between a data point x ( j) i and the cluster centre j c, is an indicator of the distance of the n data points from their respective cluster centers. The steps of the algorithm: 1. Place K points into the space represented by the objects that are being clustered. These points represent initial group centroids. 2. Assign each object to the group that has the closest centroid. 3. When all objects have been assigned, recalculate the positions of the K centroids. Repeat Steps 2 and 3 until the centroids no longer move. This produces a separation of the objects into groups from which the metric to be minimized can be calculated. F. Association Rule Mining Apriori algorithm is an example of association rule mining algorithm. Using this algorithm, candidate patterns that receive sufficient support from the database are considered for transformation into a rule. This type of algorithm works well for complete data with discrete values. A number of data mining algorithms have been introduced to the community that perform summarization of the data, classification of data with respect to a target attribute, deviation detection and other forms of data characterization and interpretation. One popular summarization and pattern extraction algorithm is the association rule algorithm, which identifies correlations between items in transactional databases [10]. Given a set of transactions, each described by an unordered set of items, an association rule X _Y may be discovered in the data, where X and Y are conjunctions of items. The intuitive meaning of such a rule is that transactions in the database, which contain the items in X, tend to also contain the items in Y. An example of such a rule might be many observed customers who purchase tires and auto accessories also buy some automotive services. In this case, X = {tires, auto accessories} and Y = {automotive services}. Two numbers are associated with each rule that indicates the support and confidence of the rule [16]. The support of the rule X _ Y represents the percentage of transactions from the original database that contain both X and Y. The confidence of rule X_Y represents the percentage of transactions containing items in X that also contain items in Y. Applications of association rule mining include cross marketing, attached mailing, catalog design and customer segmentation. An association rule discovery algorithm searches the space of all possible patterns for rules that meet the userspecified support and confidence thresholds. The problem of discovering association rules can be divided into two steps: Find all item sets whose support is greater than the specified threshold. Item sets with minimum support are called frequent item sets. Generate association rules from the frequent item sets. To do this, consider all partitioning of the item set into rule left-hand and right-hand sides. Confidence of a candidate rule X_Y is calculated as support (XY) / support (X). All rules that meet the confidence threshold are reported as discoveries of the algorithm. L1: = {frequent 1-itemsets}; k:= 2; // k represents the pass number While (Lk-1) Ck = New candidates of size k generated from Lk-1 For all transactions t D Increment count of all candidates in Ck That are contained in t Lk = All candidates in Ck with minimum support k = k+1 Report Uk Lk as the discovered frequent item sets Table-I summarizes the Apriori algorithm. The first pass of the algorithm calculates single item frequencies to determine the frequent 1-itemsets. Each subsequent pass k discovers frequent item sets of size k. To do this, the frequent item sets Lk-1 found in the previous iteration are joined to generate the candidate item sets Ck. Next; the support for candidates in Ck is calculated through one sweep of the transaction list. k-item set Lk Ck Uk Lk TABLE I APRIORI ALGORITHMS An item set containing k items Set of frequent k-item sets(k-item sets with minimum support) Set of candidate k-item sets (potentially frequent item sets) Set of generated item sets From Lk-1, the set of all frequent (k-1) item sets; the set of candidate k-item sets is created. The intuition behind this Apriori candidate generation procedure is that if an item set X has minimum support, so do all the subsets of X. Thus new item sets are created from (k-1) item sets p and q by listing p.item1, p.item2, p.item (k-1), q.item (k-1). Items p and q are selected if items 1 through k-2 are equivalent for p and q and item k-1 is not equivalent. Once candidates are generated, items etc., are removed from consideration if any (k-1) subset of the candidate is not in Lk-1. G. Weighted Rule Mining Technique The general association rule mining uses the frequency values of the attributes. The candidate sets are prepared with attribute and its values. The item sets integrates the candidate sets with its frequency values. The support and confidence values are estimated using the frequency values. The frequency based rule mining is not suitable for all types of applications [1]. Some transactions uses pre assigned weight values with its frequency value for the mining process. The stock sales application based mining uses the frequency with the profit value as weight. Some application did not have any S.Karthik et.al. 987 www.ijcsmr.org

pre assigned weight value. In this case automatic weight estimation mechanism is applied for the transactions [14]. The stock trend analysis system is focused to analyze the trade flow with trend changes. The script data values are collected for a set of companies in different sectors. The sector index and market index are associated with the script data values. The script price changes are categorized into three stages. They are positive zone, stable zone and negative zone. The rule mining is performed on the partitioned data values. The script data variations, sector index data variation and market index variations are analyzed in the association analysis phase. The profit level and price variation with index variations are used for the weight estimation process. The weight values are used in the weighted rule mining process. The weighted support is calculated with the item frequencies and weight values. In the same way the weighted confidence values are also calculated. The pattern mining process is tuned to analyze the weighted support and weighted confidence values. V. METHODOLOGY AND IMPLEMENTATION A. Integrated Clustering and Weighted Rule Mining Algorithm The stock market trend analysis is carried out with the support of the integrated clustering and weighted rule mining algorithm. This algorithm is proposed to solve the trend estimation problems. The clustering is performed on the preprocessed stock market data values. In the preprocessing stage the data values are collected from a set of web sites. The system uses three types of data values. They are script data, sector index data and market index data values. The weighted association rule mining algorithm is designed to mine attribute relationship. The attribute analysis method is designed to extract attribute relationship such as script, sector and market indexes. The system uses the frequency value in the frequent pattern mining process. The system integrates the clustering and weighted rule mining process for stock market trend analysis. The earlier cluster based association algorithm is modified with the following enhancements. The association mining uses the frequency based pattern extraction. In this case the data values and index values are associated themselves. The association mining is performed with the frequency and weight values. Market flow based clustering and association analysis replaced with market price based clustering and weight based association mining algorithm. The pattern mining is used in the prediction process. The following steps are involved in the integrated clustering and weight based rule mining models. Step-1: The stock data and index data values are fetched from the database. The system uses real data values from the stock market environment. Step-2: The preprocessing mechanism updates the stock change, sector index and market index values. Step-3: The clustering process is initiated with stock price and index data values. Step-4: The partitioning process is done with three partition levels. Step-5: Update the cluster results with partition information. K-means clustering algorithm is used in the data partitioning process. The data partitioning is done with trade price difference information. Step-6: Estimate the weight values with reference to the stock price and its change levels. Step-7: Apply the relationship mining on the partitioned data values. Step-8: Perform the weighted rule mining process for the partitioned objects by using the weight values. Step-9: Apply the trade trend estimation on the weighted rule mining results. B. Integrated Clustering and Weighted Rule Mining Technique The system is designed to estimate stock market trade trends. The share market index and company details are used in the trade trend analysis. The system imports data from the stock exchanges datasets. The trade transactions under BSE and NSE are used in the system. The system is divided into four phases. They are 1. Trade Management Phase, 2.Clustering Phase 3. Trade Relations Phase and 4. Trend Analyzer Phase Trade Management The trade management phase is designed to maintain index and company trade information. Index transaction indicates for the market and the sector. The index and company data values are collected from the NSE and BSE. The architecture of integrated clustering and weighted rule mining process is shown in Fig.1 The data values are preprocessed before the clustering and rule mining process. The index list shows the list of indexes and their information. The index details produce the daily transaction information. The stock market index shows the transaction of the entire market. The sector index shows the transactions of the specified sector only. The company list shows the list of companies traded in the stock market. The company detail shows the trade transactions of each company. Trade date, price and quantity information are listed for each company. Opening price, high price, low price and closing price details are provided in the system. Share Market Trade Analysis Trade Info Clusters Association Mining Fig.1 Architecture of integrated clustering and weighted rule mining process S.Karthik et.al. 988 www.ijcsmr.org

Clustering Process The data sets are clustered with category and quantity information. Share price and quantity is mainly considered in the clustering process. K-means clustering algorithm is used for the clustering process. Cosine distance measure is used for similarity analysis. Data values are divided with respect to the category value. The company, market index and sector index changes are considered as weights. Apriori algorithm is modified for the pattern extraction process. The system considers the relationship such as market index with company, market index with company price, company price with sector index value and sector index with market index values. Similar patterns are extracted using a threshold values. Trend Analyzer The trend analyzer phase is designed to predict the market or company trade or trend information. The trend analysis performed under the learned database shown in Fig.4. All the market indexes and sector exchange information s are analyzed under the trend analyzer module. Price or index variation details are used in the system. Trend prediction accuracy is improved by the system. The user can select any company from the list and submits the query with the date value. Trade prediction is applied for the companies under the database environment only. Market trend boundary values are displayed separately. Fig.2 Trade Transaction using Clustering Process Cluster list shows the list of cluster and its record count. A cluster detail shows the records in each cluster. The trade transaction using clustering process is shown in Fig.2. Finally processed results are stored in three different clusters. They are Positive Zone (PZ), Negative Zone (NZ) and Stable Zone (SZ). Positive zone cluster is grouped based on data value is greater than one. Negative zone cluster is grouped based on data value is less than one and Stable Zone cluster is grouped based on data value is one. Finally cluster results are passed into the rule mining process. Trade Relations The trade relations are extracted using the clustered data values for the company. The clustering process separates the company price information based on the previous day patterns. Company changes, market changes and index changes are calculated and compared in the trade relation s form. Weighted rule mining technique is used for the relationship analysis is represented in Fig.3. Fig.4 Trade Trend Analysis VI. FINDINGS AND RESULTS Findings 1. Statistical techniques are used for the market price prediction process but the results are not accurate. 2. Hybrid clustering association algorithm is also not suitable for stock market trend analysis with index dependency environment. 3. Statistical analyses are not appropriate to predict the trend analysis with index relationship. 4. The stock market data is analyzed with clustering and weighted rule mining technique with respect to the market flows, the data is clustered by K-Means algorithm. Results 1. The sample data has taken from the National Stock Exchange database. 2. Totally five sectors are taken into consideration namely NSE Index, BSE Index, Infotech Index, Automobile Index and Bank Index. Fig.3 Trade Relationship Analysis S.Karthik et.al. 989 www.ijcsmr.org

3. In that, 750 records of each Index from the date 03.01.2005 to 31.12.2007 are added in the Index list. 4. In NSE Index of trade transaction clustering process, the result of an Ashok Leyland company have a total process of 146 positive zones, 507 stable zones and 97 negative zones are calculated. 5. Similarly for Infotech Index-Infosys, 136 positive zones, 523 stable zones and 91 negative zones are calculated. 6. From this process, the trade trend of an Ashok Leyland and Infosys predicted the trend as bullish. The trade trend analysis also gives the closing level, low price and high price of the current day stock. VII. CONCLUSIONS AND FUTURE WORK Stock markets are the large-scale online market environment. Millions of trade transactions are carried out under the stock markets. Market flow is represented using the market index. The sector index is used to produce the sector flow. Company price information is maintained separately. Most of the trade analysis software s uses the statistical approach. Price is the main input data value for the system. No relationship analysis is conducted in any tool. The probability values are used in the statistical analysis methods. The share market trend analysis system is developed and tested with a set of real data values collected from the Indian stock exchange. Three years of trade data value for market index, sector index and company price are used in the system. The system operations are tested and verified real data values. Market index change and sector index changes are considered in the analysis process. All the stock market data values are analyzed with dependency factor. The relationships are used to extract trade trend patterns. There are numerous tools available in the market place which helps with the execution of predictive analytics. These range from those which need very little user sophistication to those that are designed for the expert practitioner. The difference between these tools is often in the level of customization and heavy data lifting allowed. The system produces exact price prediction for the company data. The testing results show that the trade prediction scheme produces more accurate results. The future work of this system can be enhanced with the following features. The system can be integrated with genetic algorithms for data optimizations. The frequent pattern mining model can be enhanced to perform pattern identification on data streams to analyze the live market analysis. Support and density based dissimilarity measures can be integrated in the clustering process. The system can be enhanced to perform distributed pattern mining process. REFERENCES 1. Arun K.Pujari, "Data mining Techniques" University Press, First Edition, 2001. 2. Aurangzeb Khan and Khairullah khan, Frequent Patterns Minning Of Stock Data Using Hybrid Clustering Association Algorithm 2009 International Conference on Information Management and Engineering, IEEE, 2009. 3. Chi-Lin Lu, Ta-Cheng Chen, A study of applying data mining approach to the information disclosure for Taiwan's stock market investors Expert Systems with Applications, Volume 36, Issue 2, Part 2, March 2009, Pages 3536 3542. 4. Hellstrom T. Application Of Neural Networks In Financial Data Mining, Licentiate Thesis, Department of Computing Science, Ume a University, Sweden, 1998. 5. http://www.bseindia.com/about/introbse.asp. 6. http://www.nseindia.com/content/equities/eq_scriphi stdata.htm. 7. Jar-Long Wang, Shu-Hui Chan, Stock market trading rule discovery using two-layer bias decision tree, Expert Systems with Applications, Volume 30, Issue 4, May 2006, Pages 605 611. 8. Jie S, Hui L Data mining method for listed companies financial distress prediction, Knowledge-Based Systems. 2008. 9. Jo Ting, Mining of stock data: inter-and inter-stock pattern associative classification proceedings of 2006 international conference on data mining Las Vegas, USA, 2006. 10. Kotsiantis.S, Kanellopoulos, Association Rules Mining : A Recent Overview GESTS International Transactions on Computer Science and Engineering, Vol.32(1), 2006, pp 71-82. 11. Muh-Cherng Wu, Sheng-Yu Lin, Chia-Hsin Lin, An effective application of decision tree to stock trading, Expert Systems with Applications, Volume 31, Issue 2, August 2006, Pages 270 274. 12. Senthamarai Kannan.K Sailapathi Sekar.P, Mohamed Sathik.M and Arumugam.P, Financial Stock Market Forecast using Data Mining Techniques, Proceedings of the International MultiConference of Engineers and Computer Scientists 2010. 13. Sharma, A.K and Batra G.S Indian Stock Market Publisher: Deep and Deep Publications, India 2002. 14. Soon.L.K and Sang Ho Lee, Explorative Data Mining on Stock Data Experimental Results and Findings Springer- ADMA LNAI 4632, pp. 562 569, 2007. 15. Sureshkumar.K.K. and Elango.N.M, An Efficient Approach to Forecast Indian Stock Market Price and their Performance Analysis, International Journal of Computer Applications (0975 8887) Volume 34 No.5, November 2011. 16. Ying Ma, Guanyi Li, and Zengchang Qin, Minority Game Data Mining for Stock Market Predictions Beihang University, Beijing, 2010. S.Karthik et.al. 990 www.ijcsmr.org