International Conference on Computing, Communication and Automation (ICCCA2015)

Astrological Prediction for Profession Using

Classification Techniques of Artificial Intelligence

Neelam Chaplot Praveen Dhyani

Department of Computer Science Executive Director
Banasthali University Banasthali University
Banasthali, Rajasthan, India Banasthali, Rajasthan, India
[email protected] [email protected]

O.P. Rishi
Department of Computer Science
University of Kota
Kota, Rajasthan, India
[email protected]

Abstract— Astrology has started around 4000 years back and of paper [2] reviewed applications such as medical prediction
has significantly developed over a period of time. Till date no system [3], weather forecasting[4], financial analysis [5]. The
unified rules or standards for astrological prediction exist in the techniques used in applications were Logistic regression[6],
world. Astrologers concentrate on providing quality services to Bayesian networks[7], Simple Cart algorithm[8], decision
persons rather than defining universal rules and standards for table algorithm[9], neural networks[10], case based
astrological prediction. Advances in artificial intelligence resulted reasoning[11] and so on. The results produced by these
in large number of applications for analysis and prediction. In techniques were quite good. Astrology also works on the
these applications computer learn from unknown, large, noisy or principle of classification and prediction so above methods can
complex data sets and perform prediction and classification of
be used for astrological prediction.
data. In this paper we are trying to find universal rules and
validity of astrology using various scientific methods. In this Till now hardly any large scale researches has taken place
paper we are going to predict profession of person using ZeroR, to prove validity of astrology some initial work was done in
Simple Cart and Decision Table classification algorithm. The data these field by authors of paper[12][13]. Authors designed a
set for learning classification consisted of 24 records of Singer, 24 model implement artificial intelligence for astrological
records of Player and 10 records of Scientist. Weka tool[1] prediction and nearest neighbor algorithm was developed for
available under General public license is use to perform analysis astrological prediction of profession of the person.
and prediction task.
In this paper we are going to explore supervised
Keywords—Artificial Intelligencet; Clasification; Machine classification techniques of artificial intelligence such as
Learning: Astrological Predection; Prediction of Profession; ZeroR, Simple Cart and Decision Table algorithms for
prediction of profession of a person.

I. INTRODUCTION II. CLASSIFICATION TECHNIQUES OF ARTIFICIAL

INTELLIGENCE
Astrological prediction of different incidences taking place
in human life is performed based on similarity of positions of Artificial Intelligent Techniques used for classification and
planets, zodiac signs and aspect between them. The astrologer prediction are used in variety of applications, after comparing
uses the analysis and similarity previous horoscope to perform this methods it was observed that a particular method works
prediction for person. The astrologers use similarity analysis better for one domain but not for all domains hence it is
and diagnosis of the previous horoscopes of the person to necessary to explore through all the methods to identify which
perform prediction of various events in the life of the person in method works for particular domain. Classification and
front of him/her. The profession of a person can be determined prediction methods use a variety of statistical, probabilistic and
in astrology by examination of birth chart but the rules used by optimization techniques to allow computers to learn from past
astrologers for prediction are not standardised and hardly any examples having large, noisy or complex data sets. In this
scientific proof is available. There is requirement that paper we are going to test three methods ZeroR, Simple Cart
researcher and astrologers come together and combine their algorithm and Decision table algorithm for the performance of
efforts to provide uniform rules and scientific validity for astrological prediction for profession of the person. An
prediction. Techniques in artificial intelligence are used to overview of classification techniques used in this paper is
perform classification and prediction in many fields. Authors provided below.

ISBN:978-1-4799-8890-7/15/$31.00 ©2015 IEEE 233

Authorized licensed use limited to: Lulea University of Technology. Downloaded on February 26,2024 at 06:32:45 UTC from IEEE Xplore. Restrictions apply.
 International Conference on Computing, Communication and Automation (ICCCA2015)
A. ZeroR Algorithm
ZeroR [14] is classification technique that depends on the
simple logic of working on target and ignoring its predictors.
The prediction is done by this method by predicting the
majority class and ignoring all other. It does not have much TABLE I. ATTRIBUTES USED FOR CLASSIFICATION TASKS
prediction power, it determines a baseline performance for S.No. Attribute Type Description
other classification methods. House number of Aries in the birth
1. Aries Nominal chart. Value can be a number from
B. Simple Cart Algorithm 1 to 12
House number of Taurus in the
Simple CART stands for Classification and Regression 2. Taurus Nominal birth chart. Value can be a number
Trees[8]. Cart can be applied as classification tree or as a from 1 to 12
regressive tree based on whether response variable is House number of Gemini in the
categorical or continuous. It constructs binary tree, hence each 3. Gemini Nominal birth chart. Value can be a number
from 1 to 12
internal node has exactly two outgoing edges. The splits are House number of Cancer in the
selected by identifying that which variable should be used for 4. Cancer Nominal birth chart. Value can be a number
splitting and determining precise rules for splitting [14] from 1 to 12
House number of Leo in the birth
C. Decision Table Algorithm 5. Leo Nominal chart. Value can be a number from
1 to 12
Decision Table consist of two major parts [9] first is House number of Virgo in the birth
schema i.e. list of attributes and second is the body i.e. a multi 6. Virgo Nominal chart. Value can be a number from
set of labeled instances. Set of instances with the same values 1 to 12
for schema attributes is called cell. Decision table classifier House number of Libra in the birth
7. Libra Nominal chart. Value can be a number from
based on the concept of simple lookup table. The decision 1 to 12
table majority returns the majority of the training set if the House number of Scorpio in the
decision table cell matching the new instance is empty. The 8. Scorpio Nominal birth chart. Value can be a number
decision table local searches for a decision table entry with from 1 to 12
fewer matching attributes if the matching cell is empty. This House number of Sagittarius in the
9. Sagitarius Nominal birth chart. Value can be a number
variant therefore returns an answer from the local from 1 to 12
neighborhood. House number of Capricorn in the
10. Capricorn Nominal birth chart. Value can be a number
III. PROBLEM FORMULATION from 1 to 12
House number of Aquarius in the
Classification methods can be applied on the data for 11. Aquarius Nominal birth chart. Value can be a number
astrological prediction of profession of the person. The basic from 1 to 12
steps used for problem formulation is collecting the dataset House number of Pisces in the birth
12. Pisces Nominal chart. Value can be a number from
and then identifying and isolating the features which will be 1 to 12
informative and useful. Once the dataset is ready then House number of Sun in the birth
different supervised learning methods will be used for 13. Sun Nominal chart. Value can be a number from
performing prediction. 1 to 12
House number of Moon in the birth
14. Moon Nominal chart. Value can be a number from
A. Data collection and extraction 1 to 12
Data of 24 Player, 24 Singer and 10 Scientist were House number of Mars in the birth
collected through reliable sources by the authors. The 15. Mars Nominal chart. Value can be a number from
1 to 12
information collected was Date of Birth, Time of Birth, Place House number of Mercury in the
of Birth, Profession, Education and brief biography of the 16. Mercury Nominal birth chart. Value can be a number
person. Then astrological charts were prepared based on from 1 to 12
Indian astrology. These planetary charts were constructed House number of Venus in the birth
using date of birth, time of birth and place of birth of the 17. Venus Nominal chart. Value can be a number from
person. Then these records were stored in the tabular form for 1 to 12
House number of Jupiter in the
future use. 18. Jupiter Nominal birth chart. Value can be a number
Attributes relevant to the analysis were isolated and from 1 to 12
House number of Saturn in the
collected from the above information and saved in the case 19. Saturn Nominal birth chart. Value can be a number
base. Total 23 attributes were considered and used for from 1 to 12
classification. Table 1. gives a brief description of the House number of Rahu in the birth
attributes used in the astrological prediction of profession of 20. Rahu Nominal chart. Value can be a number from
person using classification techniques. Most of the attributes 1 to 12
are nominal type holding the planetary positions of planets and House number of Ketu in the birth
21 Ketu Nominal chart. Value can be a number from
zodiacs in the horoscope at the time of birth of the person. 1 to 12

234

Authorized licensed use limited to: Lulea University of Technology. Downloaded on February 26,2024 at 06:32:45 UTC from IEEE Xplore. Restrictions apply.
 International Conference on Computing, Communication and Automation (ICCCA2015)
S.No. Attribute Type Description Decision table with 12 fold validation produced best
22. Gender Nominal
It will have value M for Male or F results for these cases. Graph in figure 1 describes the errors
for Female. generated by various methods for astrological prediction thus
It holds either of the three values providing insight into the error. Graph in figure 2 describes the
23. Class Nominal
Player, Singer and Scientist
correct and incorrect results generated by various
classification techniques on data.
B. Data Transformation 0.6
Weka (Waikato Environment for Knowledge Analysis)[1] 0.5
well known software of machine learning written in Java and
developed by University of Waikato. WEKA is software 0.4
available under the GNU General Public License. It contains a 0.3
collection of visualization tools and algorithms for data
0.2
analysis and predictive modeling, together with graphical user Mean absolute error
interfaces for easy access to this functionality [1]. 0.1
Root mean squared error
The file in ARRF format to be used by Weka tool was 0
created based on the data generated and created in data
extraction phase. The file was created by describing the
attributes and different values that those attributes can hold.
Then the records consisting of the attribute values were stored
in each line using comma separated format.

IV. EXPERIMENTS AND EVALUATIONS

Fig. 1. Chart for Comparison of the errors.

TABLE II. VALUES OF RECALL AND PRECISION

Classifier Cross Class Precision Recall
Name Validation
ZeroR 10 fold Player 0.353 0.5
ZeroR 10 fold Singer 0.333 0.333
ZeroR 10 fold Scientist 0 0
ZeroR 12 fold Player 0.414 1
ZeroR 12 fold Singer 0 0
ZeroR 12 fold Scientist 0 0
SimpleCart 10 fold Player 0.25 0.25
SimpleCart 10 fold Singer 0.323 0.417
SimpleCart 10 fold Scientist 0.333 0.1
SimpleCart 12 fold Player 0.372 0.667
SimpleCart 12 fold Singer 0.231 0.125
SimpleCart 12 fold Scientist 0.5 0.1
Fig. 2. Chart for Comparison of Correct and Incorrect Classification.
Decision 10 fold Player 0.457 0.667
Table
Decision 10 fold Singer 0.522 0.5
Table
V. CONCLUSIONS AND FUTURE WORK
Decision 10 fold Scientist 0 0 With the results produced in fig 2. It can be concluded that
Table decision table algorithm with 12 fold cross validation is better
Decision 12 fold Player 0.472 0.708
Table
for predicting profession of person than other algorithms
Decision 12 fold Singer 0.545 0.5 discussed because it produced best results with 50% accuracy.
Table
Even though the accuracy of result produced is not very
Decision 12 fold Scientist 0 0
Table high. But as our data set consisted of only 58 cases so by
increasing number of cases the accuracy of the result will
Data created in the above process was loaded in the Weka increase and as only few of the classification techniques are
tool. Then various classification techniques such as ZeroR, tested hence there is a requirement to test other prediction and
Simple Cart and Decision Table algorithms were applied on classification techniques. Also as we are working on different
the data with 10 and 12 fold cross validation. Table 2. Gives classifier we are able to identify the weakness and strength of
the values of precision and recall produced by the various different methods so we can develop a classification method
methods. Fig 1. presents the comparison of errors generated which utilizes the strength of various classifiers and thus in
by methods. Fig 2. Presents graph of correct and incorrect turn increases the accuracy in prediction. Also different pair of
classification produced by various classification techniques.

235

Authorized licensed use limited to: Lulea University of Technology. Downloaded on February 26,2024 at 06:32:45 UTC from IEEE Xplore. Restrictions apply.
 International Conference on Computing, Communication and Automation (ICCCA2015)
attributes can be identified and selected which may produce predicting pneumonia mortality in Artificial Intelligence in Medicine”,
more relevant results. Elsevier, 1997, vol. 9, pp. 107- 138.
[4] FolorunshoOlaiya, Adesesan Barnabas Adeyemo, “Application of
We have performed prediction for profession of the
Data Mining Techniques in weather prediction and climate change
person by using different classification techniques similarly
studies”, I.J. Information Engineering and Electronic Business,
we can predict the basic nature of the person, the attitude of
2012, vol. 1, pp. 51-59.
the person, money, family and other aspect of human life.
[5] Pei-Chann Chang, Chen-Hao Liu, Jun-Lin Lin, Chin-Yuan Fan, Celeste
S.P. Ng, “A neural network with a case based dynamic window for stock
ACKNOWLEDGMENT trading prediction”, Expert Systems Applications, Elsevier, April 2009,
Vol 36(3). pp. 6889-6898.
It is not possible to carry forward a research work without
[6] Le Cessie, S., Van Houwelingen, J.C., “Ridge Estimators in Logistic
any help, we would like to express our sincere appreciation and Regression”, Applied Statistics,1992, 41(1), pp191-201.
gratitude to all the person who have directly or indirectly [7] Jensen F., An introduction to Bayesian networks, Springer,1996.
helped us in these research. We would also like to express [8] Leo Breiman, Jerome H. Friedman, Richard A. Olshen, Charles J. Stone,
special thanks to the people who have provided their personal “Classification and Regression Trees”, Wadsworth International Group,
details to carry forward this research. Belmont California, 1984.
[9] Ron Kohavi and Daniel Sommer, “Targeting Business Users with
Decision Table Classifiers”, KDD-98 Proceedings. American
REFERENCES Association of Artificial Intelligence, 1998.
[1] G. Mark Hall, Eibe Frank, Geoffrey Holmes, Bernhard Pfahringer, Peter [10] Zheng, Z., “Constructing X-of-N Attributes for Decision Tree
Reutemann, and Ian H. Witten, “The WEKA data mining software”, Learning”, Machine Learning, 2000, vol. 40, pp. 35–75.
ACM SIGKDD Explorations Newsletter, vol. 11(1), pp. 10-18, 2009. [11] Waston and FarhiMarir ,”Case Based Reasoning: A Review”,The
Clerk Maxwell, A Treatise on Electricity and Magnetism, 3rd ed., vol. 2. Knowledge Engineering Review, 1994, Vol.9 , pp. 4.
Oxford: Clarendon, 1892, pp.68-73.
O.P. Rishi and Neelam Chaplot, “Predictive role of case based reasoning
[2] Neelam Chaplot , Praveen Dhyani , O.P.Rishi, “A Review on Machine for astrological predictions about profession: System modeling
Learning Concepts for Prediction Based Application”, International approach”, International Conference on Communication and
Journal of Computational Science, Engineering & Technology, ISSN Computational Intelligence, Dec 2010, pp 313 – 317.
2320-4648, March 2013, Vol 1(2), pp. 12-18. [12] O.P. Rishi and Neelam Chaplot, “Archetype of astrological prediction
system about profession of any persons' using case based reasoning” ,
[3] Gregory F. Coopera, Constantin F. Aliferis, Richard Ambrosinoa,
International Conference on Communication and Computational
John Aronisb, Bruce G. Buchananb, Richard Caruana,Michael J. Intelligence, Dec 2010, PP. 373 – 377,Dec 2010.
Fined, Clark Glymour, Geoffrey Gordon, Barbara H. Hanusad, Janine [13] Lior Rokach, “Decision Trees, Data Mining and Knowledge Discovery
E. Janoskyf, Christopher Meek, Tom Mitchell,Thomas Richardson, Handbook”, pp. 165-192.
Peter Spirtes, “An evaluation of machine-learning methods for

236

Authorized licensed use limited to: Lulea University of Technology. Downloaded on February 26,2024 at 06:32:45 UTC from IEEE Xplore. Restrictions apply.