See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/281995140

Astrological Prediction for Profession Doctor using Classification Techniques

of Artificial Intelligence

Article in International Journal of Computer Applications · July 2015

DOI: 10.5120/21778-5052

CITATIONS READS

5 2,749

3 authors, including:

Neelam Chaplot Praveen Dhyani

Manipal University Jaipur 41 PUBLICATIONS 202 CITATIONS
15 PUBLICATIONS 161 CITATIONS
SEE PROFILE
SEE PROFILE

All content following this page was uploaded by Neelam Chaplot on 31 March 2016.

The user has requested enhancement of the downloaded file.

 International Journal of Computer Applications (0975 – 8887)
Volume 122 – No.15, July 2015

Astrological Prediction for Profession Doctor using

Classification Techniques of Artificial Intelligence

Neelam Chaplot Praveen Dhyani O. P. Rishi

Department of Computer Executive Director Department of Computer
Science & Engineering Banasthali University Science & Informatics
Banasthali University Banasthali, Rajasthan, India University of Kota
Banasthali, Rajasthan, India Kota, Rajasthan, India

ABSTRACT 2. LITERATURE REVIEW

Astrology can be an interesting example of the application of Influences of planetary influences are plainly visible such as
various classification techniques of artificial intelligence. In day and night, change of seasons and tidal effects. Planets and
astrology, predictions about different aspects of human life stars surely have influence on everyday life of humans.
are done based on the planetary position of the stars at the Astrology predicts the course of human life based on the
time of birth of a person. In this research work, the positions planetary position of the stars at the time of birth of the
of the planets and stars at the time of the birth of a person are person. There must be some substance in astrology for it has
utilized. This information is used to predict the possibility of managed to survive through generations. A lot of literature is
person to become doctor. available on astrology but the scientific validity could not be
proved. Although after 15th century, lot of research has been
Total 102 records were collected for the study out of that half done to establish a scientific validity for astrology.
of the records were of persons that were doctor and other half
records of the persons that were not doctor by profession. Some researchers believe that astrology is not scientific [2][3]
Thereafter, various supervised classification techniques such while other feel that in depth study in this field is required to
as Logistic, Naïve Bayes, Simple Cart, Decision Stump, reach that conclusion[4]. There are beliefs that predictive and
Decision Table and DTNB algorithm were used and results non-predictive are two parts of astrology and that predictive
were compared for their accuracy. astrology is the proper subject for testing whether astrology
can be used to make prediction[5].
General Terms
Artificial Intelligence, Astrology, Classification Technique. Supervised classification techniques in artificial
intelligence[6] is recently developed branch which performs
Keywords well for various classification and prediction tasks. It
Artificial Intelligence, Astrological Prediction, Classification performs well for very vague and uncertain data also. These
Techniques, Prediction of Profession Doctor, Weka. techniques have been used for various applications such as
weather forecasting[7], cancer analysis and diagnostics[8],
1. INTRODUCTION machine translation[9] etc. Brief review of various machine
Astrology is the area that generates lots of curiosity among learning methods and the applications is available in
human beings. To find scientific theories and its relations with paper[10].
astrology, changes are to be done in the way astrology is By applying different learning techniques on large amount of
looked upon. To find scientific basis to astrology changes are data available for the planetary position of the persons and
to be made from the traditional ways in which astrology is their biographies, a system which can predict on various
analyzed. In this paper, efforts have been made to identify the aspects of human life can be developed.
scientific basis using various classification techniques of
artificial intelligence. These techniques are used to perform Hardly any work has been done till now in predicting the
classification of data automatically. Various classification incidences of human life based on planetary positions using
techniques used here have resemblance with the way artificial intelligence techniques. Some preliminary work
astrology works. using Case Based Reasoning and nearest neighbor algorithm
has been done in literature [11][12].
First step for prediction involves performing various
classification techniques of artificial intelligence that are 3. CLASSIFICATION TECHNIQUES
treated as black box for already existing methods which Classification techniques of artificial intelligence helps to
provided good results in various fields for classifying data. create generalized hypothesis from a set of data so that
Next step involves finding out the attributes that helps in prediction for the future can be made based on the hypothesis
improving the dynamic prediction in astrology and then generated. These techniques involve developing algorithm
updating the knowledge of the system when more records based on prior data in which class labels are known
come in picture. The results in the form of rules will be beforehand. Algorithms are typically developed on a training
generated by the classification techniques. These rules dataset and then tested on an independent test data set to
generated by classification techniques will be backtracked and evaluate the accuracy of algorithms.
verified against the rules used by astrologers. Resemblance
between these rules generated by classification techniques and It is very difficult to identify the suitable classification
rules provided by astrology will provide the scientific basis techniques for given data because some techniques perform
for astrology. well on some kind of dataset but vice versa on others. This

28
 International Journal of Computer Applications (0975 – 8887)
Volume 122 – No.15, July 2015

study investigates six classification techniques. Logistic schema. Decision table classifier is based conceptually on a
Regression, Simple Cart, Naïve Bayes Decision Stump, simple lookup table. If no matching instances are found than
Decision Table and DTNB. The selection was based on the decision table returns the majority class of the training set
difference in the techniques for learning and level of otherwise majority class of all matching instances is returned.
complexity. To build a Decision Table Majority (DTM), induction
algorithm must decide which features to include in the schema
3.1 Logistic Regression and which instances to store in the body [17].
Logistic Regression[13] is a probabilistic statistical
classification model. It is used to predict the class label based 3.5 Decision Stump Algorithm
on one or more predictor variables, these variables are not Decision Stump [18] is one-level decision tree. It is a decision
continuous. tree with one root node connected to leaf node. A decision
stump makes a prediction based on the value of just a single
The formula can be stated as the probability of occurrence of input feature hence are also called 1-rules.
an event Y =1
Based on input features, different types of trees are created.
For nominal features, a decision tree which contains a leaf for
each possible feature value is constructed or a decision tree
with two leafs is created. One leaf corresponds to some
Ln is natural logarithm and is line regression chosen category and the other leaf for all other categories. For
equation continuous features, usually, some threshold feature value is
selected, and the decision tree containing two leafs are created
Expected probability that Y = 1 for a given value of X. for values below and above the threshold.

3.6 DTNB Algorithm

DTNB algorithm [19] is a simple Bayesian network in which
decision table represents a conditional probability table.
3.2 Simple Cart Algorithm Learning algorithm splits the attributes into two disjoint
It is one of the decision tree algorithms. It does the attribute subsets, one for decision table and the other for Naive Bayes.
selection, based on implementing minimal cost-complexity Forward selection method is used, at each step selected
pruning. It constructs binary search tree. For more information attributes are modeled by Naïve Bayes and remainder by
see [14]. decision table and all attributes are modeled by decision table
initially.
CART stands for classification and regression trees. Cart can
be applied as classification tree or as a regressive tree based The class probability estimates of the Decision Table and
on whether response variable is categorical or continuous. It Naïve Bayes must be combined to generate overall class
constructs binary trees hence each internal node has exactly probability estimates. Assuming X┬ is the set of attributes in
two outgoing edges. The splits are selected by identifying that the Decision Table and X┴ the one in Naive Bayes, the
which variable should be used for splitting and determining overall class probability is computed as
precise rules for splitting[3]

3.3 Naïve Bayes Classification Algorithm

Naïve Bayes classifier is a classification algorithm based on Where and are the class
probabilities. It is based on the concept of independence of the probability estimates obtained from the Decision Table and
variable that is a Naive Bayes classifier assumes that the Naive Bayes respectively, α is a normalization constant and
presence or absence of features of a class is unrelated to the Q(y) is prior probability of class.
presence or absence of any other features.
Naive Bayes classifiers can be trained very efficiently in
4. METHDOLOGY
a supervised learning environment. In many practical Data of 102 persons were collected out of that 51 records
applications, parameter estimation for Naive Bayes models were of doctor and 51 records were of persons those are not
uses the method of maximum likelihood. In other words, one doctor by profession. Data were collected from reliable local
can work with the Naive Bayes model without believing in persons on paper and through emails. These persons were
Bayesian probability or using any Bayesian methods. from a certain geographical area of India. The information
collected was Date of Birth, Time of Birth, Place of Birth,
Let there be ‘m’ Classes: C1, C2, C3…Cm Profession and brief biography of the person.
Naïve assumes that class is conditionally independent [15] Data relevant to the analysis was generated and retrieved by
finding the planetary positions of zodiac and planets based on
the details of place of birth, date of birth and time of birth of
the person. Then these records were stored in tabular form for
future use.
The attributes used by classification techniques of artificial
intelligence are as follows.
Table 1. Attributes used for performing classification.
3.4 Decision Table Algorithm S.No. Attribute Type Value
Decision Table consist of two major parts [16]. First is
schema i.e. lists of attributes and second is the body i.e. 1 Aries Nominal Value from 1 to 12
labeled instances from the space described by the features in

29
 International Journal of Computer Applications (0975 – 8887)
Volume 122 – No.15, July 2015

2 Taurus Nominal Value from 1 to 12 techniques such as Simple Cart, Logistic, Naïve Bayes,
Decision Stump, Decision Table and DTNB were used on the
Gemini Nominal Value from 1 to 12 dataset with 10, 12, and 14 fold cross validation. The results
3
of various techniques are shown through graph in Fig. 1 for
Cancer Nominal Value from 1 to 12 correctly and incorrectly classified data for identification of
4 profession doctor of a person.
5 Leo Nominal Value from 1 to 12 Better results were produced by Simple Logistic with 12 fold
cross validation with an accuracy of 54.902 %. Decision
6 Virgo Nominal Value from 1 to 12 Stump algorithm with 14 fold classification gave results with
an accuracy of 50%.
7 Libra Nominal Value from 1 to 12
6. CONCLUSION AND FUTURE WORK
Scorpio Nominal Value from 1 to 12 With limited number of sample cases and different training
8 methods the prediction that person will become a doctor was
made. The methods that gave better results are Simple
9 Sagittarius Nominal Value from 1 to 12
Logistic with 12 fold cross validation with an accuracy of
54.902%. Decision Stump algorithm with 14 fold
10 Capricorn Nominal Value from 1 to 12 classification gave results with an accuracy of 50%.

Aquarius Nominal Value from 1 to 12 Although, the results are not as per expectations but are not
11
discouraging also. Few methods are performing better than
Pisces Nominal Value from 1 to 12 others so if remaining methods are tried or hybrid methods are
12 created than accuracy of the system can be increased. Also
data set consisted of only 102 records so by increasing the
13 Sun Nominal Value from 1 to 12
number of cases the accuracy of the results are expected to
increase further. Results generated by working on different
14 Moon Nominal Value from 1 to 12 classifiers can be used to analyze and identify the weakness
and strength of different classification methods so that new
15 Mars Nominal Value from 1 to 12 classifiers can be developed which utilizes the strength of
various classifiers and thus in turn increases the accuracy in
16 Mercury Nominal Value from 1 to 12 prediction. For the accuracy purpose, other aspects of person
information like knowledge level, family background etc.
17 Venus Nominal Value from 1 to 12 must be included as attributes along with birth charts,
astrological charts, and planetary positions in different houses
Jupiter Nominal Value from 1 to 12 in horoscope. Few more attributes like strength of the planets,
18
aspect ratio of the planets and interrelationship among planets
Saturn Nominal Value from 1 to 12 can also be included for increasing the accuracy of results.
19
Predictions are performed to find that a person will become
20 Rahu Nominal Value from 1 to 12 doctor or not similarly predictions can be performed to
identify basic nature of the person, the attitude of the person,
21 Ketu Nominal Value from 1 to 12 money and other aspect of person’s life.

Gender Nominal 7. ACKNOWLEDGMENTS

22 Male or Female
Research work of this scale is not possible without direct and
Class Nominal indirect support of many persons. We would like to thank
23 Doctor and Non Doctor them all. Also, we extend our special thanks to the persons
who have provided us with personal details to carry forward
our research. Lastly, we thank family and friends to provide
5. EXPERIMENTS AND EVALUATIONS unconditional moral support to carry forward the research.
Data created in the above process was loaded in the Weka tool
by creating the file in ARRF format. Various classification

30
 International Journal of Computer Applications (0975 – 8887)
Volume 122 – No.15, July 2015

Fig 1: Graph for Classification of Profession Doctor using various Supervised Classification Techniques

8. REFERENCES [11] O.P. Rishi and Neelam Chaplot. Dec 2010. Predictive
[1] Hall, M., Frank, E., Holmes G., Pfahringer, B., role of case based reasoning for astrological predictions
Reutemann, P., & Witten, I. H. 2009. The WEKA data about profession: System modeling approach,
mining software. ACM SIGKDD Explorations International Conference on Communication and
Newsletter. Computational Intelligence.

[2] Ivan W. Kelly. 1997. A Concept of Modern Astrology a [12] O.P. Rishi and Neelam Chaplot. Dec 2010. Archetype of
Critique. Article in Psychological Reports. astrological prediction system about profession of any
persons' using case based reasoning, in International
[3] John H. Mcgrew and Richard M. Mcfall. 1990. A Conference on Communication and Computational
Scientific Inquiry Into the Validity of Astrology. Journal Intelligence.
of Scientific Exploration.
[13] le Cessie S., Van Houw ngen J.C. 1992. Ridge
[4] Ken McRitchie. August 2011. Support for Astrology Estimators in Logistic Regression. Applied Statistics.
from the Carlson Double-blind Experiment. ISAR
International Astrologer. [14] Leo Breiman, Jerome H. Friedman, Richard A. Olshen,
Charles J. Stone. 1984. Classification and Regression
[5] Penny Seator. 2008/2009 Astrological Prediction and Trees. Wadsworth International Group, Belmont,
Statistical Tests. The British Astrological Association. California.
[6] S. B. Kotsiantis, I. D. Zaharakis, P. E. Pintelas. 2007. [15] Harry Zhang, Jiang Su. 2004. Naive Bayesian Classifiers
Machine learning: a review of classification and for Ranking. Lecture Notes in Computer Science Volume
combining techniques. Springer Science Business Media 3201.
[7] B Folorunsho Olaiya, Adesesan Barnabas Adeyemo. [16] Ron Kohavi and Daniel Sommer.1998. Targeting
2012. Application of Data Mining Techniques in Business Users with Decision Table Classifiers. KDD-98
Weather Prediction and Climate Change Studies. I. J. Proceedings American Association of Artificial
Information Engineering and Electronic Business. Intelligence.
[8] Joseph A. Cruz and David S. Wishart. 2006. [17] Kohavi R, 1995. The Power of Decision Tables. In
Applications of Machine Learning in cancer prediction Proceeding European Conference on Machine Learning.
and prognosis cancer.Informatica
[18] Iba, Wayne and Langley, Pat. July 1992. Induction of
[9] Harold Somers. June 1999. Review Article: Example- One-Level Decision Trees. ML92: Proceedings of the
based Machine Translation. Machine Translation. Ninth International Conference on Machine Learning,
Aberdeen, Scotland, San Francisco, CA, Morgan
[10] Neelam Chaplot, Praveen Dhyani, O.P.Rishi. March
Kaufmann.
2013. A Review on Machine Learning Concepts for
Prediction Based Application, International Journal of [19] Mark Hall, Eibe Frank. 2008. Combining Naive Bayes
Computational Science, Engineering & Technology. and Decision Tables. In Proceedings of the 21st Florida
Artificial Intelligence Society Conference.

IJCATM : www.ijcaonline.org
31

View publication stats