IRJET- Titanic Survival Analysis using Logistic Regression

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 06 Issue: 08 | Aug 2019 www.irjet.net p-ISSN: 2395-0072
© 2019, IRJET | Impact Factor value: 7.34 | ISO 9001:2008 Certified Journal | Page 89
Titanic Survival Analysis using Logistic Regression
Vaishnav Kshirsagar1, Nahush Phalke2
1Graduate student, University of San Francisco, California, USA
2Software Engineer, Accenture, Pune, Maharashtra, India
---------------------------------------------------------------------***----------------------------------------------------------------------
Abstract - The sinking of the Titanic ship caused the death
of about thousands of passengers and crew is one of the fatal
accidents in history. The loss of lives was mostly caused due to
the shortage of the life boats. The mind shaking observation
came out from the incident is that some people were more
sustainable to endure than many others, like children, women
were the one who got the more priority to be rescued. The
main objective of the algorithm is to firstly find predictable or
previously unknown data by implementing exploratory data
analytics on the available training data and then apply
different machine learning models and classifiers to complete
the analysis. This will predict which people are more likely to
survive. After this the result of applying machine learning
algorithm is analyzed on the basis of performance and
accuracy.
Key Words: Logistic Regression, Data Analysis, Kaggle
Titanic Dataset, Data pre-processing. Cross validation,
Confusion Matrix
1. INTRODUCTION
The field of machine learning has allowed analysts to
uncover insights from historical data andpastevents. Titanic
disaster is one of the most famous shipwrecks in the world
history. Titanic is a British ship liner that sank in the North
Atlantic Ocean, a few hours after colliding with an iceberg.
While there are facts available to support the cause of the
incident of ship breaking, there are various speculations
regarding the survival rate of passengers in the Titanic
disaster. Over the years, data of survived as well asdeceased
passengers has been collected. The dataset is publicly
available on a website called Kaggle.com.
This dataset has been studied and analyzed using
various machine learning algorithms like Random Forest,
SVM etc. Various languages and tools are used to implement
these algorithms including Weka, Python, R, Java etc. The
approach of the research paper is centered on R and Python
for executing algorithms- Nave Bayes, Logistic Regression,
Decision Tree, and Random Forest. The prime objective of
the research is to analyze Titanic disaster to determine a
correlation between the survival of passengers and
characteristics of the passengers using various machine
learning algorithms. In particular, this research work
compares the algorithms on the basis of the percentage of
accuracy on a test dataset.
2. ALGORITHM
2.1 Data Pre-processing
In the dataset available for the prediction some of the data
values are missing or unknown. This missing data was
resulting in reducing the accuracy of the overall prediction
model and also reduces the size of pure training data which
in turn reduces accuracy. Data preprocessing is a technique
that involves transforming raw data into an understandable
format.
Real-world data is often incomplete, inconsistent, and/or
lacking in certain behaviors or trends,andislikelytocontain
many errors. Data preprocessing is a proven method of
resolving such issues. Data preprocessingpreparesrawdata
for further processing. Missing values are replaced by
average of that column. So, the missing and unknowndata of
the passengers which is easily predictable is filled up by this
step.
2.2 Classification
Logistic Regression:
Second step of the algorithm is using a classifier to classify
the available information. Logistic Regression is the
appropriate regression analysis to conduct when the
dependent variable is dichotomous (binary). Like all
regression analyses, the logistic regression is a predictive
analysis.

Logistic regression is used to describe data and to explain
the relationship between one dependent binaryvariable and
one or more nominal, ordinal, interval or ratio-level
independent variables. It uses a method of using he
regression line between dependent and independent
variable to predict the value of the dependent variable.
2.3 Cross validation
Dataset is divided into two mainpartsnamely Trainand Test
data. Training data will be considered for the training of the
machine. Test data will be used for validating the machine.
Cross validation technique used here is K-Fold.
The method has only one parameter called k that refers to
the number of groups into which a given data sampleistobe
split. As such, the method is also called k-fold cross-
validation. When a particular value for k is chosen, it may be
used in place of k in the reference to the model, such as k=10
becoming 10-fold cross-validation.
2.4 Analysis of confusion matrix
Confusion matrix is used to show the performance of the
algorithm. Accuracy of the model can be predicted using the
confusion matrix. It is a plotting of relation betweenreal and
predicted outputs. It allows us to check the accuracy and
performance of the algorithm. In this case we are using two
attributes at a time for the confusion matrix plotting. Test
case data is used to build the confusion matrix.
The values shown in the confusion matrix are the
probability of survival of the individual considering only
those parameters. As shown in fig [2] the cell on firstcolumn
and is of age and the 7th row is sex_male i.e. the probability
of surviving the individual is depending on the age and the
gender as if he is male is 0.081. As it is positive there is a
possibility that the person with this attribute survives.
3. RESULTS
The logistic regression gives the accuracy of 95% which is
based on the confusion matrix. The parameters used here
are accuracy and false discovery rate. Accuracy is a measure
of the correctness of the prediction of the model. Higher
accuracy is always better and is calculated by
(TN + TP)/Total number of rows *100
False discovery rate are the false positive measures
of confusion matrix where the model predicts that the
passenger would survive but in reality,itdoesn’t. This would
prove dangerous as the prediction may go wrong and
hampers the accuracy of the results. The attempts are being
made to increase the accuracy rate and reduce the false
discovery rates.
4. CONCLUSIONS
The logistic regression provides a better accuracyi.e.almost
of about 95%. It works better with binary dependent
variable which means the variable has a binary value as its
output like yes or no, true or false.

Fig 2: Confusion Matrix
The ROC curve is the plotting of the outputbasedonthefalse
positive rate and the true positive rate plotted along x and
the y-axes. The Curve depicts the performance of various
algorithms on the same data which helps to compare the
performance, accuracy and efficiency of the algorithm. It
helps to decide the best algorithmwhichissuitableforuser’s
requirement.
REFERENCES
[1] Analyzing Titanic disaster using machine learning
algorithms-Computing,CommunicationandAutomation
(ICCCA), 2017International Conferenceon21December
2017, IEEE.
[2] Prediction of Survivors in Titanic Dataset: A
Comparative Study using Machine LearningAlgorithms,
Tryambak Chatterlee, IJERMT-2017.
[3] MICHAEL AARON WHITLEY, using statistical learningto
predict survival of passengers on the RMS Titanic by
Michael Aaron Whitley, 2015.
[4] Lonnie Stevans, David L. Gleicher,” Who Survived the
Titanic? A logistic regression analysis”-Article in
International Journal of Maritime History, December
2004.
[5] MICHAEL AARON WHITLEY, using statistical learningto
predict survival of passengers on the RMS Titanic by
Michael Aaron Whitley, 2015.
[6] Bircan H., Logistic Regression Analysis: Practice in
Medical Data,KocaeliUniversitySocial SciencesInstitute
Journal, 2004 / 2: 185- 208
[7] Atakurt, Y., 1999, Logistic Regression Analysis and an
Implementation in Its Use in Medicine, Ankara
University Faculty of Medicine Journal, C.52, Issue 4,
P.195, Ankara
[8] Kaggle, Titanic: Machine Learning form Disaster
[Online]. Available: http://www.kaggle.com/

IRJET- Titanic Survival Analysis using Logistic Regression

More Related Content

What's hot (19)

Similar to IRJET- Titanic Survival Analysis using Logistic Regression (20)

More from IRJET Journal (20)

Recently uploaded (20)

IRJET- Titanic Survival Analysis using Logistic Regression