Predictve data mining

1. Data Mining Methods &
Implementation of Predictive
Data Mining Architecture
Research Scholar
VENKATA RAMANA KATHI
Research Supervisor
Dr. GEETANJALI AMARAWAT
Professor & HOD Dept. of CSE
Madhav University

2. What is Prediction
• What you think will happen in the Feature
• Predictive data mining makes prediction about
unknown data values by using the known
values via classification, regression, time
series analysis, prediction, etc
Two types of Predictions
• Stock market Prediction
• Environment Prediction
Introduction

3. Stock Market Prediction
• This thesis are to augment the forecast capacity of financial
stock market utilizing time arrangement quantitative technical
investigation
• Stock market prediction is the act of trying to determine the
future value of a company stock or other financial instrument
traded on an exchange.
• The successful prediction of a stock's future price
could yield significant profit.
• This basic part examines the exploration issues in anticipating
the stock market.

4. Various parameters yield in Stock Market
• To devise a remarkable metric in light of "Rectify Profitable
Trade Signal-CPTS" for assessment of the stock market
foreseeing algorithms.
• To examine the execution of the Moving Average Crossover
(MAC) Algorithm on the noteworthy time arrangement data.
• The rate of yearly Returns on Investment (ROI) produced.

5. Problem Specification
• This exploration concentrates on the effective Market Basket
Analysis system to give the most obliged things to the client,
in view of their buy conduct.
• In Stock market prediction taken as the problem for
prediction
• Stock market prediction is the act of trying to determine the
future value of a company stock.
• The successful prediction of a stock's future price could yield
significant profit.
• This research to derive an algorithm which is to build a
accuracy & Time series to find the future prediction pattern.

6. Various types of stock market exchanges in India
• Bombay Stock Exchange
• National Stock Exchange
• Regional Stock Exchanges
• Ahmedabad Stock Exchange
• Bangalore Stock Exchange
• Bhubaneshwar Stock Exchange
• Calcutta Stock Exchange
• Cochin Stock Exchange
• Coimbatore Stock Exchange
• Delhi Stock Exchange

7. Financial Market Analysis
• An economy which relies primarily on interactions between
buyers and sellers to allocate resources is known as a market
economy.
• The financial markets can be divided into different subtypes:
1. Capital markets which consist of Stock markets, which
provide financing through the issuance of shares or common
stock.
Research Analysis

8. 2. Commodity markets, which facilitate the trading of
commodities.
3. Money markets, which provide short term debt financing and
investment.
4. Derivatives markets, which provide instruments for the
management of financial risk.
5. Insurance markets, which facilitate the redistribution of various
risks.
• All these markets are very unsafe and require considerable
information and experience to counteract substantial misfortunes.
• They also require close attention to market developments.
• Stocks, then again, are less hazardous because developments of the
market are usually gradual.

9. Data Mining & Analysis
• Data mining is a solitary stride in a larger procedure of
Knowledge Discovery in Databases (KDD).
• KDD is thought to be an all the more encompassing procedure
that incorporates data warehousing, target data choice.
• Data cleaning, pre-processing, transformation and lessening,
data mining, show determination, evaluation and interpretation,
and finally consolidation and utilization of the extracted
"learning".

10. Descriptive Data Mining
• This enables to see sets of summarized data in brief, distinct
terms.
• An idea usually alludes to a gathering of data, for example,
stereos, visit purchasers, graduate understudies, and so on
• Characterization gives a compact and concise summarization of
the given accumulation of data, while idea or class comparison,
also known as discrimination gives portrayals comparing at
least two accumulations of data.

11. Predictive Mining
• It is an analytic procedure intended to investigate large
amounts of data in search of predictable patterns and/or
systematic relationships amongst variables, and then to
validate the discoveries by applying the identified patterns to
new subsets of data.
• It is utilized to forecast express values, based on patterns
decided from known outcomes
• At the point when market beating strategies are found via data
mining, there are various potential issues in making the leap
from a back-tried strategy to effectively putting resources into
future real world conditions.

12. • The primary issue is deciding the probability that the
relationships are not random at all market conditions.
• to varying conditions and affirming that the time arrangement
patterns have statistically significant prescient power for the
accompanying two potential reasons.
• 1. High probability of profitable trades and
• 2. High profitable returns for the investment.

13. Research Methodology
• The ultimate goal is expectation – and prescient data mining is
the most widely recognized kind of data mining and one that has
the most direct business applications.
• This is finished utilizing large noteworthy market data to speak to
varying conditions and affirming that the time arrangement
patterns have statistically significant prescient power for the
accompanying two potential reasons.
1. High probability of profitable trades and
2. High profitable returns for the investment.

14. Algorithm
• An "efficient" market is characterized as a market where there are
huge quantities of objective, "benefit maximizes“ effectively
contending.
The Efficient Market Hypothesis (EMH)
• The EMH proposes two urgent ideas that have characterized the
discussion on efficient markets from that point.
1. Types of informational efficiency.
2. Joint Hypothesis problem.
1. Informational Efficiency
• A market is called educational efficient if prices dependably filly reflect
accessible data.

15. • Enlightening effectiveness of stock market prices proposed by EMH is
one of the foundations of modem financial hypothesis.
• The following three elements appear in the production of
information.
1. The normal return - if the normal profit of exchanging based
for the data is "too low", at that point the data may not be
delivered.
2. The cost of delivering data with which money must be earned.
3. The cost of scattering tile data, with the end goal that the
normal return for sure will be figured it out.
• The sort of data that will be created and the speed of joining of the
data in prices rely upon the over three components.

16. The EMH recognized three noteworthy types of market
proficiency:
• Weak Form Efficiency- Prices completely reflect data with respect
to the past successions of prices. No financial specialist can
acquire overabundance returns by creating exchanging rules
construct exclusively with respect to authentic price or return data.
• Semi-Strong Form Efficiency - Prices completely mirror all freely
accessible data, including financial proclamation data. No financial
specialist can acquire overabundance returns firm utilizing
exchanging rules in view of any openly accessible data.
• Solid Form Efficiency - Prices completely mirror all data,
including inside data. No financial specialist can acquire
abundance returns utilizing any data whether openly accessible or
not.

17. • Joint Hypothesis Problem
The idea of "Joint Hypothesis Problem" in EMH is shown as
the thought of market effectiveness, which couldn't be
rejected without a going with dismissal of the model of
market harmony.
The advantage evaluating model depends on the model of
market harmony and it is a circumstance where the supply of
anything is precisely equivalent to its demand.
• Fundamental Analysis
Fundamental Analysis depends on the macroeconomic data
and the essential financial status of organizations like money
supply, loan fee, inflationary rates, profit yields,

18. Top-Down Approach
• The best down contributing takes a gander at an organization's
working condition not with standing its own techniques and
likely future execution.
• The systems can be compared to a rearranged pyramid.
• Adopting a best down strategy to an organization's profit's
prospects includes first taking a gander at the expansive
macroeconomic, social and political condition in which the
organization works.

19. Bottom-Up Approach
• The Bottom-up essential analysis channels the diverse
organizations in a segment by taking a gander at the individual
"venture story" of each, and breaking down a significant
number of various numbers.
• These numbers incorporate both the financial articulations
distributed by each organization and also particular proportions
that speculators ascertain from these in a procedure that is
known as "calculating".
• In the event that the "inherent esteem" is lower than the
overarching share price, expert rates the stock as a "purchase";
on the off chance that it is lower, at that point the examiner
suggest an "offer".

20. RESULTS AND ANALYSIS
• Number of profitable trade signals generated,
• Percentage of CPTS (Correct Profitable Trade Signal)
generated
• Percentage of false or non-productive signals generated
and
• Percentage of annual ROI of money.
The Performance Evaluation Metrics

21. • The goal of this proposition, "Advancement of Predictive Data
Mining Architecture", is to plan and create calculations to gauge
the stock market incline in light of the entomb day monetary
time series data.
• To plan and create calculations to gauge the stock market
incline in light of the entomb day monetary time series data.
• All the monetary time series anticipating calculations revealed
till date utilize "Era of Correct Trade Signal - GCTS" metric
mirroring the Market Trend to demonstrate the effectiveness of
the particular calculations.
• 'Rates of return - ROT, to be specific, ensured profit is a 'Prime
Metric' to quantify the "Effectiveness" of any monetary time
series gauging calculation.

22. Profitability Calculation
• When number of equities is traded over a period of time on the
historic data, a trade consolidation needs the figure of profit / loss
to be calculated.
• Each buy or sell transaction has its own cost involved in the
transaction called as "Buy Cost" or "Sell Cost" respectively.
• Cost of single trade = ("Sell Cost) - ("Buy Cost")
• Positive results indicate a profitable trade and negative results
indicate a loss. The percentage of ROI is calculated as given in the
equation 6.2. ROI % = (((Sell Cost) - (Buy Cost)) / (Buy Cost)) * 100

23. • The minus percent denotes the percentage of loss and plus
percent denotes the percentage of profit obtained in the
respective trade of a stock.
• The sum of all cost committed for buying stocks over the
period under study gives the "Total Buy Cost".
• Net = (Total Sell Cost) - (Total Buy Cost),
• If the Net is negative, it is a loss and if the Net is positive it is
a profit. This is the Net Profit / Loss.

24. Results
Fig 1: Various Stock list

25. Performance Comparison In Total Number of Trade
Signals

26. Graphical view of the comparative performance in
terms of total number of trade signals generated

27. Graphical view of the comparative performance in terms of
total number of CPTS generated ( Correct Profitable Trade
Signal)

28. Graphical view of the comparative performance in
terms of percentage CPTS generated

29. Graphical view of the comparative performance in
terms of percentage of non productive signals generated

30. CONCLUSION
• The primary concentration of this proposition was to build up a novel
class of combinational quantitative specialized analytics to foresee the
share trading system for comprehending certain pivotal issues in
money related securities exchange prediction distinguished amid
writing study, specifically, era of conspicuous purchase/offer trade
signals yielding higher rate of yearly profits for the venture;
recognizing and joining all around situated specialized markers with
the end goal of producing "Correct Profit Trade Signals - CPTS", a
theoretical metric presented in this examination work yielding shut
trade signals. Developing Effective Market Hypothesis (EMH) changing
wasteful machine learning calculations into powerful calculations
instantly grabbing genuine noticeable purchase and offer signals for
preparing however they happened to be less overwhelming.

31. References
1. Abe, M. “Theories and model building for CRM data analysis RF analysis based on consumer behavior theories.”
Ryutsu oho (Distribution Information), December, 2004. (in Japanese)
2. Agrawal, R. Imielinski, T. and Swami, A. “Mining Association Rules Between Sets of Items in large Databases.”
Proceedings of the 1993 ACM- SIGMOD International Conference on Management of Data, Washington DC, USA,
May 1993. 207-216.
3. Ailawadi, Kusum L., Scott A. Neslin, amd Karen Gedenk. “Pursuing the Value-Conscious Consumer: Store Brands
Versus National Brand Promotions.” JM, 65.1
4. Anand, S. S. Bell, D. A. Hughes, J.G. “EDM: A General Framework for Data Mining Based on Evidence Theory.” Data
and Knowledge Engineering Journal18.3 (1996): 189- 223.
5. Anand, S.S. Patrick, A.R. Hughes, J.G and Bell, D.A. “A Data Mining Methodology for Cross Sales.” Journal of
Knowledge-Based Systems 10.7 (1998): 449-461.
6. Anon. “Data mining efforts increase business productivity and
efficiency.”2001.www.bettermanagement.com/library/library.aspx?libr aryid=5664. Online.
7. Anon. Introduction to data mining and knowledge discovery. Third edition. Potomac, MD: Two Crows Corporation,
1999. Print.
8. Apte C. Liu B. Pednault E.P.D and Smyth P. “Business Applications of Data Mining.” Communications of the ACM,
Special Issue: Evolving data mining into solutions for insights 45.8 (2002): 49-53.
9. Bart, Lariviere. Dirk, Van. and Den, Poel. “Predicting Customer Retention and Profitability by Using Random Forests
and Regression Forests techniques.” Expert Systems with Applications 29 (2005): 472-494.

32. • Bartlett M.S., “Test of Significance for Factor Analysis” 103-375-765 (Last edited on 2002/04/17 10:36:30 GMT-6), 1950.
• Baumgartner, Hans and Steenkamp, Jan-Benedict E. M. “Exploratory Consumer Buying Behavior: Conceptualization and
Measurement.” International Journal of Research in Marketing. 13.2 (1996): 121-137.
• Berman, Barry and Evans, Joel R. Retail Management A Strategic Approach. New Delhi: Pearson Education, 2004. Print.
• Berry M.J.A. and Linoff G.S. Mastering Data Mining: The Art and Science ofCustomer Relationship Management. Canada: John
Wiley & Sons, 2000. Print.
• Berson A. Smith S. and Thearling K. Building data mining applications for CRM. McGraw-Hill, 2000.
• Bhasin Madan Lal. “Data Mining: A Competitive Tool in Banking and Retail Industry.” The Chartered Accountant, October 2006.
• Bose I. and Mahapatra R.K. “Business Data Mining - A Machine Learning Perspective.” Information and Management 39 (2001):
211-225.
• Bounsaythip Catherine and Rinta-Runsala Esa. “Overview of Data Mining for Customer Behavior Modeling.” VTT Information
Technology Research report TTEI- 2001.
• Brand, E. and Gerritsen, R. Data Mining Solutions Supplement. White Paper from Exclusive Ore Inc., 1998. http://www.xore.com.
Online.
• Buchanan, B.G. Brief History of Artificial Intelligence. 2006 <http://www.aaai.org/AITopics/bbhist.html>.
• Buck N. “Eureka! Knowledge Discovery.” Software Magazine 20 (2001): 24-29.
• Buckinx Wouter, Versttraeten Geert, Poel Dirk Van Den. “Predicting Customer Loyalty Using the Internet Transactional Database.”
Expert Systems with Applications 32 (2007): 125-134.
• Bukowitz, W.R. and Williams, R.L. The Knowledge Management Fieldboook. Edinburgh: Pearson, 1999. Print.
• Cahlink, George. “Data Mining Taps the Trends,” Government Executive Magazine, October1, 2000.
<http://www.govexec.com/tech/articles/1000managetech. htm.>
• Carrier C. G. and Povel O. “Characterizing Data Mining Software.” Intelligent Data Analysis 7(2003): 181–192.