AN

INTERNSHIP PROJECT REPORT ON

Titanic Passengers’ Survival Analysis
A Project Report for Internship Programme
Submitted by
ANKITA BASAK
ANINDYA GIRI
ARITRA DATTA
SHEETAL GUPTA

in partial fulfillment for the award of the degree of

Business Analytics
in
MBA

At

Euphoria GenX

1|Page
 DECLARATION

I undersigned, hereby declare that the project titled “Titanic Passengers’

Survival Analysis” submitted in partial fulfillment for the award of Degree of
Master of Business Administration of MAKAUT is a bonafide record of work
done by me under the guidance of Animesh Ojha. This report has not previously
formed the basis for the award of any degree, diploma, or similar title of any
University.

14/10/2021 Anindya Giri

2|Page
 BONAFIDE CERTIFICATE

Certified that this project work was carried out under my supervision
Development of a feature-rich, Titanic Passengers’ Survival Analysis
is the bonafide work of

Student Name Signature

ANINDYA GIRI

ANKITA BASAK

SHEETAL GUPTA

ARITRA DATTA

----------------------------------
SIGNATURE

Name :
ANIMESH OJHA

3|Page
 MCKVIE
MBA in Business Analytics

CERTIFICATE

This is to certify that the report titled ‘Titanic Passengers’ Survival Analysis’
being submitted by Anindya Giri, (11610120001) in partial fulfilment of the
requirements for the award of the Degree of Master of Business Administration,
is a bonafide record of the project work done by Anindya Giri ,Ankita Basak,
Aritra Datta and Sheetal Gupta of department of MBA, MCKVIE.

Name of guide – ANIMESH OJHA

Designation -
Director -

4|Page
 ACKNOWLEDGEMENT

Through this acknowledgement we express our sincere gratitude towards all those
people who helped us in this project, which has been a learning experience.
This space wouldn’t be enough to extend our warm gratitude towards our project
guide Mr. Animesh Ojha for his efforts in coordinating with our work and
guiding in right direction.
It would be injustice to proceed without acknowledging those vital supports We
received from our beloved classmates and friends, without whom would have
been half done.
We would like to thank Mr. Diptayan Bhattacheryya, for helping us to get such
an excellent opportunity.
We also use this space to offer our sincere love to our parents and all others who
had been there, helping us walk through this work.

Anindya Giri
Ankita Basak
Aritra Datta
Sheetal Gupta

5|Page
 List of Tables
Table Title of the Table Page
No. No.

1 Abstract 10

2 Introduction 11

3 Research Methodology 15

4 Dataset (Titanic Passengers’ Survival Analysis) 18

5 Data Analysis 23

6 Findings & Discussion 56

7 Implementation 58

8 Conclusions 59

9 References 60

6|Page
 List of Figures

Fig. No. Title of the Figure Page

No.

1 RESEARCH DESIGN FLOWCHART 16

2 COUNTPLOT FOR SURVIVED AND NOT SURVIVED 25

3 COUNTPLOT FOR SURVIVED, SEX, PCLASS, EMBARKED 26

4 PLOTTING AVERAGE SURVIVAL RATE BASED ON SEX 27

5 PLOTTING DISTPLOT FOR AGE AND FARE 28

6 PLOTTING DISTPLOT FOR FARE AFTER NORMALIZATION AND 29

STANDARDIZATION
7 PLOTTING COUNTPLOT OF DIFFERENT PORT OF EMBARKATION 39
WITH RESPECT TO ‘PCLASS’
8 PLOTTING NORMALIZED CONFUSION MATRIX 49

9 LEARNING CURVE 53

7|Page
 List of Symbols/ Abbreviations
Symbol/ Explanation Pag
Abbreviation e
No.
pd import pandas as pd (for data loding, cleaning, analysis and 24
manipulation)

np import numpy as np (for mathematical Calculations) 24

plt import matplotlib.pyplot as plt (data visualization) 24

sns import seaborn as sns (data visualization) 24

ax Axis 28

ss StandardScaler 29

rb RobustScaler 29

mm MinMaxScaler 29

qcut Quantile-cut 41

gs GridSearchCV 44

8|Page
 TABLE OF CONTENTS

Page
Sl No. Topics
No
1 ABSTRACT 10
2 INTRODUCTION 11
2.1 BACKGROUND OF THE STUDY 12
2.2 NEED AND SIGNIFICANCE OF THE STUDY 13
2.3 STATEMENT OF PROBLEM 13
2.4 OBJECTIVES OF THE STUDY 13
2.5 SCOPE OF THE STUDY 13
2.6 LIMITATIONS OF THE STUDY 14
2.7 ORGANIZATION OF THE REPORT 14
3 RESEARCH METHODOLOGY 15
3.1 OBJECTIVES 15
3.2 HYPOTHESIS 15
3.3 RESEARCH DESIGN 16
3.4 SOFTWARE REQUIREMENT 17
3.5 TOOLS DESCRIPTION 17
4 DATASET (Titanic Passengers’ Survival Analysis) 18
4.1 SOURCES OF DATA 18
4.2 PRIMARY AND SECONDARY DATA 19
4.3 POPULATION 20
4.4 SAMPLE DESIGN 20
4.5 SAMPLING METHOD 20
4.6 METHOD OF DATA COLLECTION 21
4.7 DRAFTING A QUESTIONNAIRE 21
4.8 DATA ANALYSIS TECHNIQUES 22
5 DATA ANALYSIS 23
5.1 MACHINE LEARNING PROJECT: TITANIC PASSENGERS’ 24
SURVIVAL ANALYSIS
5.2 25
VISULIZATION & ANALYSIS
5.3 HYPERPARAMETER TUNING 44
5.4 MODEL ESTIMATION AND EVALUATION 45
5.5 CONFUSION MATRIX 47
5.6 CLASSIFICATION REPORT 50
5.7 LEARNING CURVE 52
5.8 PREDICTIONS FOR TEST DATASET 54
6 FINDINGS (DISCUSSION) 56
7 58
IMPLEMENTATION
8 CONCLUSIONS 59
9 REFERENCES 60

9|Page
 CHAPTER-1
ABSTRACT

With the greater development of technology and automation human history is

predominantly updated. The technology movement shifted from large mainframes to PCs to
cloud when computing the available data for a larger period. This has happened only due to
the advent of many tools and practices, that elevated the next generation in computing. A
large number of techniques has been developed so far to automate such computing. Research
dragged towards training the computers to behave similar to human intelligence. Here the
diversity of machine learning came into play for knowledge discovery. Machine Learning (ML)
is applied in many areas such as medical, marketing, telecommunications, and stock, health
care and so on. This paper presents reviews about machine learning algorithm foundations, its
types and flavours together with Python scripts possibly for each machine learning
techniques.

10 | P a g e
 CHAPTER-2
INTRODUCTION

ML denotes to the methods tangled in distributing through massive facts in the greatest
intellectual way to arise better understandings. ML algorithms are defined to be culturing an
objective function (f) which better draws input identifier (g) to an output identifier (h) as in
equation,h = f(g)
This future output prediction is not that much easier to do manually. Hence an automated
system is expected to do the process ]. Thus use of machine learning algorithms come into the
scene. For every new input (g) the output (h) is predicted genuinely using machine learning
algorithms. This state is said to be predictive molding/predictive analytics. The major operation
is to assess he most possible predictions with the present data. Each data is segregated as
training set and testing set as in figure

Five basic steps for a ML task:

1. Data accumulation: Data gathered from various sources are used for analysis.

2. Data preprocessing: Before getting into the actual processing of data, preprocessing is
mandatory. This step is used to noise or other unwanted data from the gathered data.

3. Prototype training: This step contains selecting the suitable algorithm and depiction of data
in a pattern (model) format. The preprocessed data is often divided into two parts namely
training and testing data.

4. Pattern evaluation: In this step, the resultant pattern is validated for its correctness. However
substantial time is required in data gathering and training.

5. Performance enrichment: This step involves picking another different pattern with better
efficiency.

Despite any model/pattern, the above said five steps are mandatory to configure the ML
technique.
Types of machine learning algorithms–scenario based

Supervised learning (SL) or prognostic models

This is used to assess the upcoming result with the help of chronological data . These models
are instructive as much concentration is emphasized in training phase . For instance, SL is
applied if a selling firm wishes to find its customers list. It could also be used in prediction of
earthquakes, cyclones etc. to determine the Insurance credit. Few examples of these prediction
algorithms are: multiple linear regression logistic regression NaiveBayes, DecisionTrees

Unsupervised learning (UL) or evocative models UL

It is suitable to train vivid models with no target and no sole feature is significant compared to
one another . For instance, UL is applied in case if a vender desires to find which product does
the customer buys frequently. Moreover, in medicinal business, UL may be applied to envisage

11 | P a g e
the diseases that may prone to occur laterally with diabetes. Few examples of UL based
algorithms are, Simple K- means clustering

Reinforcement learning (RL)

RL is applied when the system is trained to yield decisions automatically with the business
requirements only with a sole motto to exploit better effectiveness (performance) . The
underlying idea is a software agent is trained in an environment for problem solving. This
repeated learning procedure promises lower human proficiency thus saving human effort [6].
One best example of RL algorithm is Markov Decision Process

The goal of the project was to predict the survival of passengers based off a setof data. We
used Kaggle competition "Titanic: Machine Learning from Disaster" (see
https://www.kaggle.com/c/titanic/data) to retrieve necessary dataand evaluate accuracy of
our predictions. The historical data has been split intotwo groups, a 'training set' and a 'test
set'. For the training set, we are provided with the outcome (whether or not a passenger
survived). We used this set to build our model to generate predictions for the test set. For
each passenger in the test set, we had to predict whether or not they survived the sinking.
Our score was the percentage of correct predictions.

Work Plan

1) Learn programming language Python

2) Learn to use SciKit-Learn library in Python, including a Using LinearRegressi-on

algorithm and classification algorithm.

3) Performing Feature Engineering, applying machine learning algorithms, andanalyzing

results

2.1 BACKGROUND OF THE STUDY

Titanic disaster occurred about 100 years back but still it attracts the researchers to
understand and study that how some passengers survived and others perished. In this work,
the characteristics of the passengers will be identified and the relationship of survival chance
from the disaster is found. Feature engineering techniques will be performed, where the
alphabetic values will be changed to numeric values, the family size will be calculated. Also,
we will extract the title from the name, and deck label from ticket number. Classification is
done using Decision tree machine learning classification algorithm using two classes which
are survived and not survived. R programming has been used for its implementation.
Clustering is performed using KMeans machine learning algorithm. Its implementation has
been done using Python programming.

12 | P a g e
 2.2 NEED AND SIGNIFICANCE OF THE STUDY

The Titanic was competitively designed to be the best, but failed--the sea winning the battle.
The sinking of the Titanic in the early hours of April 15, 1912, remains the
quintessentialdisaster of the century. A total of 1,517 souls--men, women and children - lost
their lives ( only 711 survived). The Titanic sunk because it was built to be the best, and
thus caused people in charge to be full of pride, which caused the Marconi telegraph
distress signals to beignored. This analysis is been done so that we can learn from the
Titanic’s failure and analysethe causes and effects of the disaster.

2.3 STATEMENT OF THE PROBLEM

The RMS Titanic, a luxury steamship, sank in the early hours of April 15, 1912, off the
coast of Newfoundland in the North Atlantic after sideswiping an iceberg duringits maiden
voyage. Of the 2,240 passengers and crew on board, more than 1,500
lost their lives in the disaster. There was a huge loss which passengers faced during Titanic
disaster. To come out from this type of problem the case study of analysis onTitanic disaster
dataset is needed.

2.4 OBJECTIVE OF THE STUDY

Feature Engineering
Since the data can have missing fields, incomplete fields, or fields containing hidden
information, a crucial step in building any prediction system is Feature Engineering. For
instance, the fields Age, Fare, and Embarked in the training and test data, had missing values
that had to be filled in. The field Name while being useless itself, contained passenger's Title
(Mr., Mrs., etc.), we also used passenger's surname to distinguish families on board of
Titanic. Below is the list of all changes that has been made to the data- Extracting Title from
Name. In this project, final destiny of all people will be predicted given explanatory
variables that cover approx. all aspects of travelers. The goal of this project is to create a
classification model that is able to accurately estimate the survival of the people in the ship.
We can use this model if any same kind of accident happen in future to predict the survival
rate.

2.5 SCOPE OF THE STUDY

The study aims to analyze the number of people survived, gender and survival rate. The goal
of this project is to create a regression model that is able to accurately estimate the loses
occurred during Titanic disaster.

13 | P a g e
 2.6 LIMITATION OF THE STUDY

There is no guarantee that the data will be available in time nor contains the exact requested
list of features. Thus, there might be a risk that the access will be deniedor delayed. If so, the
study will be accomplished based only on the public dataset. Moreover, this study will not
cover all regression algorithms. Also the data collected is too old.

2.7 ORGANIZATION OF THE REPORT

The dataset in this section is the so-called Titanic dataset. It contains data concerning survival
analysis of passengers.

14 | P a g e
 CHAPTER-3
RESEARCH METHODOLOGY

3.1 OBJECTIVES:

The objective is to perform exploratory data analytics on various information in the dataset
available and to know effect of each field on survival of passengers by applying analytics
between every field of dataset with “Survival” field. The predictions are done for newer data
sets by applying machine learning algorithm. The data analysis will be done on applied
algorithms and accuracy will be checked. Different algorithms are compared on the basis of
accuracy and the best performing model is suggested for predictions. The aim of this study is to
get as reliable results as possible from the raw and missing data by using machine learning and
feature engineering methods. In the test data, all the attributes of train data will be mentioned
except the survived column. So, the objective is to get the survival status of each passenger in
the test data.
The goal of this research paper is to correctly predict who would survive the Titanic given a set
of demographic information. A predictive model using passenger data was built so people’s
genders, their ages, what class of ticket they belonged from, and their socio-economic class, all
contributed to whether they would be lucky enough to survive or tragically perish on the
Titanic.

3.2 HYPOTHESIS:

Obtaining valuable results from the raw and missing data by using machine learning and feature
engineering methods is very important for knowledge-based world. In this paper, we have proposed
models for predicting whether a person survived the Titanic disaster or not. The accuracy of the model is
evaluated using “confusion matrix”. A confusion matrix is a table layout that allows to visualize the
correctness and the performance of an algorithm. The dataset found on the Kaggle website has two
perspectives website has two perspectives. One is a training data and the other is a testing data. The
objective of the training data is to create a model which will help in predicting the outcomes of the test
data. For the purpose of this research paper, the training data from the Kaggle website will be divided
into two parts using three different ratios for training and creating the model and then predicting and the
testing data from the Kaggle website will not be used. There will be application of
different techniques for predicting whether a person survived the Titanic disaster or not. The data
analysis will then be done and the prediction outcomes will be checked for accuracy. The accuracy will
then be compared in order to suggest the better performing algorithm with respect to used dataset.
With respective to the Titanic dataset study, we have a 2 different studies of research hypothesis.
1) Simple Hypothesis: - the study of the relationship between a single dependent variable and a
single independent variable visualizes correctly and make better sense to calculate the outliers and
threshold
2) Complex hypothesis: - In our Titanic dataset it predicts the relationship between two or more
independent and dependent variables correctly by having different threshold value.

15 | P a g e
 3.3 RESEARCH DESIGN:

To pick a specific model, there is a step-by-step technique. It is important to determine if the

model to be implemented is appropriate for the given issue. In the below diagram, flow of
processes can be seen acting in accordance here.

Feature engineering is the most important part of data analytics process. It deals with, selecting
the features that are used in training and making predictions. In feature engineering the domain
knowledge is used to find features in the dataset which are helpful in building machine learning
model. It helps in understanding the dataset in terms of modelling. A bad feature selection may
lead to less accurate or poor predictive model. The accuracy and the predictive power depend
on the choice of correct features. It filters out all the unused or redundant features.

16 | P a g e
 3.4 SOFTWARE REQUIREMENT

This project uses the following tools for its implementation:

1) Python: the programming language used to implement this project

2) Jupyter notebook: used to provide an interactive environment for python, for the
implementation of this project

3) Scikit-Learn: used to implement the several machine learning models used in this project as
as view their accuracies.

4) Matplotlib: used to plot graphs for us to understand the trend in accuracy/performance of the
various machine learning algorithm implemented.

5) Seaborn: use for visualization of the statistical behaviour of data set.

6) Pandas: - used to analyse and clean our dataset,

7) Numpy: for mathematical Calculations.

3.5 TOOLS DESCRIPTION

The various tools used in this project are described below.

1) Python: Python is an interpreted, high level programming language created by Guido ban
Rossum in 1991 it emphasizes on code readability by using whitespace to terminate statements
and blocks. It ease of use and syntax makes it a language of preference for data analysis

2) Jupyter Notebook: It is web application that provides an integrated development

environment for python using this one can share documents that contains equation,
visualizations such as graphs test as a live for this reason it is highly used tool for the purposes
of data analysis

3) Scikit-Learn: It is python package used for data modelling. It provides a number of

supervised and unsupervised machine learning models. Scikit-learn makes it extremely simple
to train models with simple function calls on the input data being all that is needed

4) Matplotlib: Matplotlib is a cross-platform, data visualization and graphical plotting library

for Python and its numerical extension

5) Pandas: Pandas is an open-source Python package that is most widely used for data
science/data analysis and machine learning tasks. It is built on top of another package named
NumPy, which provides support for multi-dimensional arrays .

6) Numpy: NumPy can be used to perform a wide variety of mathematical operations on arrays.
It adds powerful data structures to Python that guarantee efficient calculations with arrays and
matrices and it supplies an enormous library of high-level mathematical functions that operate
on these arrays and matrices
17 | P a g e
 CHAPTER 4
DATASET (TITANIC PASSENGERS’ SURVIVAL
ANALYSIS)

4.1 SOURCES OF DATA:

The following is a representation of the test dataset provided in a comma separated value (CSV)
format from Kaggle and 891 rows of data (a subset of the entire passenger manifest). The
Titanic dataset consist of a training set that includes 891 passengers and a test set that includes
418 passengers which are different from the passengers in training set. A description of the
features is given in TableI.

18 | P a g e
Data Description

VARIABLE DESCRIPTIONS:

• pclass Passenger Class

o (1 = 1st; 2 = 2nd; 3 = 3rd)
• survival Survival
o (0 = No; 1 = Yes)
• name Name
• sex Sex
• age Age
• sibsp Number of Siblings/Spouses Aboard
• parch Number of Parents/Children Aboard
• ticket Ticket Number
• fare Passenger Fare
• cabin Cabin
• embarked Port of Embarkation
(C = Cherbourg; Q = Queenstown; S = Southampton)
• home.dest Home/Destination

• SPECIAL NOTES:
Pclass is a proxy for socio-economic status (SES)
1st ~ Upper; 2nd ~ Middle; 3rd ~ Lower

Age is in Years; Fractional if Age less than One (1)

If the Age is Estimated, it is in the form xx.5

With respect to the family relation variables (i.e. sibsp and parch)
some relations were ignored. The following are the definitions used
for sibsp and parch.

Sibling: Brother, Sister, Stepbrother, or Stepsister of Passenger Aboard Titanic

Spouse: Husband or Wife of Passenger Aboard Titanic (Mistresses and Fiances Ignored)
Parent: Mother or Father of Passenger Aboard Titanic
Child: Son, Daughter, Stepson, or Stepdaughter of Passenger Aboard Titanic

Other family relatives excluded from this study include cousins,

nephews/nieces, aunts/uncles, and in-laws. Some children travelled
only with a nanny, therefore parch=0 for them. As well, some
travelled with very close friends or neighbors in a village, however,
the definitions do not support such relations.

4.2 PRIMARY AND SECONDARY DATA:

The data we used for our study was provided on the Kaggle website. We were given 891
passenger samples for our training set and their associated labels of whether or not the
passenger survived.

For each passenger, we were given his/her passenger class, name, sex, age, number of
siblings/spouses aboard, number of parents/children aboard, ticket number, fare, cabin
embarked, and port of embarkation (as shown in Table 1). For the test data, we had 418
samples in the same format. The dataset is not complete, meaning that for several samples, one
19 | P a g e
 or many of fields were not available and marked empty (especially in the latter fields – age,
fare, cabin, and port). However, all sample points contained at least information about gender
and passenger class. After importing the packages and libraries to make writing the program
simpler, the data from the seaborn package is needed to be loaded and thereby few rows are to
be printed. The data was then required to be analyzed by receiving counts of it. Following the
receipt of a count of the dataset’s rows and column, it is helpful to keep a record of every rows
of the passengers who were aboard the ship. On the other hand, the columns are regarded as
every commuter’s assigned data. Theoretically, there were 891 passengers/rows and 15 data
points/columns in the data set. Meanwhile, statistics like mean, count, standard deviation, etc.
was received on the dataset.

4.3 POPULATION:
Based on data set obtained from Kaggle, the purpose is about survival prediction of the titanic. We will
split the historical data set that we obtained from kaggle into two groups – train set and test set. They are
the two different files in csv format. The train data will be an input for the training model. This data will
be used to build the model for the generations of predictions for the test data. There can be many flaws
in the train set because of which training of the model on the basis of such data is not possible or can
leads to an incorrect output. It is possible that in the csv file of the train set, there will be few blank cells
where no values is mentioned. For an example, if the age of a person is not mentioned then this a flaw in
the train data set. Hence, it has to be filled with appropriately calculated age value before training phase.
This is feature engineering that we do to make the train data set perfect for giving it as an input for
training the model.

4.4 SAMPLE DESIGN:

Clustering the data based upon classifications and use of clustering analysis simple associations
may be understood from the data. While an association might be strong through this analysis,
the true cause and effect cannot be concluded. Similarly, there are modifications that needs to
be performed on training data set. The prediction that whether or not the titanic passengers
survived the sinking will be found for each passenger in the test set. Different machine learning
techniques will be applied on the data set available. By the results obtained from those
techniques we will be able to understand the statistics more clearly. Algorithms used includes
feature engineering, decision tree visualization, random forest etc.

4.5 SAMPLING METHOD:

Feature Engineering is required for modifying the train data set because there can be some empty spaces,
title for some passengers will be missing, department for some crew members may not be mentioned.
Visualization of relationship between the family size and survival is required. Hence for all such errors
this technique has be applied before applying the machine learning techniques for training the model.
The best way for doing this using programming for data extraction and analyzing using graphs.
Classification Technique is used to determine that for which class or category a given observation will
belong to. There are many algorithms related to classification but for this work, the decision tree
algorithm is employed. All nodes in the tree will indicate either the passenger survived or not survived
using 0 and 1 respectively. At the first level, the relation between the child nodes and the root node will
indicate the gender of passenger. Further, the relation will be based on age, Passenger class and family
20 | P a g e
size. Accuracy for the resulted survival status in the test data has to be calculated by comparing with the
survival status of passengers in train data. Clustering Technique is used to group a set of observations in
such a way that observation of similar types belongs to a group. There can be many groups defined
while clustering. This technique has also to be implemented for examining passenger’s survival in titanic
disaster.

4.6 METHOD OF DATA COLLECTION:

Before applying any type of data analytics on the dataset, the data is first cleaned. There are some
missing values in the dataset which needs to be handled. In attributes like Age, Cabin and Embarked,
missing values are replaced with random sample from existing age. The data is first cleaned before
implementing some form of data analytics on the dataset. Here, it needs to be analyzed. It is also seen
that there are a few absent data values for the column: age as it is smaller than 891. Absent values are
restored with arbitrary samples. Using the age of a person, their sex, title, cabin number, Pclass, place of
embarkation, fare and size of family which includes both parch as well as sibsp, the following features
on the exploratory study is done. The survival column is selected as the reaction column. Such
characteristics are chosen because their ideals have an effect on the survival rate. If wrong features are
chosen, bad predictions may be generated. Thus, in creating an effective predictive model, function
engineering works as a backbone. NaN, NAN and na are some of the columns that contain empty values.
Only four columns, namely age, deck, embarked and embarked_town have a few absent values. In an
attempt to display a few unwanted columns, the next step is to have a look every column’s value name
as well as count. The non-redundant rows and columns should be dropped, and their rows and missing
values should be removed.

4.7 DRAFTING A QUESTIONNAIRE

Over the years, a lot of thought has been put into the science of the design of survey questions.
Key design principles:

1. Keep the questionnaire as short as possible.

2. Ask short, simple, and clearly worded questions.

3. Start with demographic questions to help respondents get started comfortably.

4. Use dichotomous (yes | no) and multiple-choice questions.

5. Use open-ended questions cautiously.

6. Avoid using leading-questions.

7. Pretest a questionnaire on a small number of people.

8. Think about the way you intend to use the collected data when preparing the questionnaire.

21 | P a g e
 4.8 DATA ANALYSIS TECHNIQUES

I. Importing all the Required modules / libraries.

II. Data Loading

III. Data Cleaning

IV. Data Mining

V. Data Visualization.

VI. Splitting the Data

VII. Linear Regression Methods.

22 | P a g e
 CHAPTER-5
DATA ANALYSIS

5.1 MACHINE LEARNING PROJECT: TITANIC PASSENGERS’

SURVIVAL ANALYSIS

RMS Titanic

The RMS Titanic was a British passenger liner that sank in the North Atlantic Ocean in the
early morning hours of 15 April 1912, after it collided with an iceberg during its maiden voyage
from Southampton to New York City. There were an estimated 2,224 passengers and crew
aboard the ship, and more than 1,500 died, making it one of the deadliest commercial peacetime
maritime disasters in modern history. The RMS Titanic was the largest ship afloat at the time it
entered service and was the second of three Olympic-class ocean liners operated by the White
Star Line. The Titanic was built by the Harland and Wolff shipyard in Belfast. Thomas
Andrews, her architect, died in the disaster.

In this project, we will develop and evaluate the performance and the predictive power of a
model trained and tested on data collected from Titanic dataset.

23 | P a g e
Once we get a good fit, we will use this model to predict the chances of survival of passenger at
the Titanic.
A model like this would be very valuable for analysis of past data like Titanic dataset.

# IMPORT LIBRARIES NECESSARY FOR THIS PROJECT:

# IMPORTING THE TITANIC DATASET:

# SUMMARY OF TRAINING DATASET:

24 | P a g e
# DESCRIPTION OF THE TRAINING DATASET:

# SHAPE OF DATASET:

5.2 VISULIZATION & ANALYSIS

# COUNTPLOT FOR SURVIVED AND NOT SURVIVED:

0= NOT SURVIVED, 1= SURVIVED

25 | P a g e
In this plot we can see that number of survived is less than number of not survived.

# IDENTIFYING THE UNIQUE NUMBER OF VALUES IN THE

DATASET:

# COUNTPLOT FOR SURVIVED, SEX, PCLASS, EMBARKED:

26 | P a g e
From this plot ,
we can see that number of male passengers is greater than that of female.
Pclass = 3 has maximum number of passengers.
In case of Port of Embarkation(C = Cherbourg; Q = Queenstown; S = Southampton), S has
highest number of passengers.

# FINDING AVERAGE SURVIVAL RATE BASED ON SEX:

Here,
We can see that female has higher survival rate than that of male.

# PLOTTING AVERAGE SURVIVAL RATE BASED ON SEX:

27 | P a g e
# PLOTTING DISTPLOT FOR AGE AND FARE:

Normal distribution, also known as the Gaussian distribution, is a probability distribution that
is symmetric about the mean, showing that data near the mean are more frequent in occurrence
than data far from the mean. In graph form, normal distribution will appear as a bell curve. If
one tail is longer than another, the distribution is skewed. These distributions are sometimes
called asymmetric or asymmetrical distributions as they don’t show any kind of symmetry.
Symmetry means that one half of the distribution is a mirror image of the other half. For
example, the normal distribution is a symmetric distribution with no skew. The tails are exactly
the same.

A left-skewed(negatively skewed) distribution has a long left tail. Left-skewed distributions are
also called negatively-skewed distributions. That’s because there is a long tail in the negative
direction on the number line. The mean is also to the left of the peak.

A right-skewed(positively skewed) distribution has a long right tail. Right-skewed distributions

are also called positive-skew distributions. That’s because there is a long tail in the positive
direction on the number line. The mean is also to the right of the peak.

28 | P a g e
Both the distplots are showing normal distribution and positively skewed distribution.

# PLOTTING DISTPLOT FOR FARE AFTER NORMALIZATION AND

STANDARDIZATION:

By using RobustScaler(), we can remove the outliers and then use either StandardScaler or
MinMaxScaler for preprocessing the dataset.
Standardizing a dataset involves rescaling the distribution of values so that the mean of
observed values is 0 and the standard deviation is 1. Data normalization is the organization of
data to appear similar across all records and fields.
It increases the cohesion of entry types leading to cleansing, lead generation, segmentation, and
higher quality data.
Numpy log1p() is a mathematical function that helps the user to calculate the natural
logarithmic value of x+1 where x belongs to all the input array elements. log1p is reverse of
exp(x) – 1. So the distplot of ‘Fare’ of train dataset becomes more prominent.

29 | P a g e
# FINDING TITLE OF ALL THE PEOPLE INSIDE TITANIC:

# FINDING AVERAGE SURVIVAL RATE WITH RESPECT TO TITLE:

Here,
We can see that ‘Lady’,’Mlle’, ‘Mme’, ‘ Ms’, ‘Sir’, ‘the’ – these titles have highest survival
rate.

30 | P a g e
# FINDING THE AVERAGE SURVIVAL RATE WITH RESPEST TO
LENGTH OF NAME:

The pandas documentation describes qcut as a “Quantile-based discretization function.” This

basically means that qcut tries to divide up the underlying data into equal sized bins. The
function defines the bins using percentiles based on the distribution of the data, not the actual
numeric edges of the bins. Here the ‘Name_Len’ of train dataset is divided into 5 bins to find
the average survival rate of each bin.

# FINDING NUMBER OF NAMES IN DIFFERENT NAME LENGTH

BINS:

# FINDING NUMBER OF PEOPLE IN DIFFERENT SEX:

31 | P a g e
# FINDING AVERAGE SURVIVAL RATE WITH RESPECT TO SEX:

# FINDING AVERAGE SURVIVAL RATE WITH RESPECT TO AGE

(NULL OR NOT NULL):

#FINDING AVERAGE SURVIVAL RATE WITH RESPECT TO

DIFFERENT AGE GROUP:

Here,
We can see that Age group 0.419 to 19.0 has highest survival rate. Age group 19.0 to 25.0 has
lowest survival rate.

# FINDING NUMBER OF PEOPLE IN DIFFERENT AGE GROUP:

32 | P a g e
Here,
We can see that Age group 0.419 to 19.0 has highest number of people. Age group 25.0 to 31.0
has lowest number of people.

#FINDING AVERAGE SURVIVAL RATE WITH RESPECT TO NUMBER

OF SIBLINGS/SPOUSES ABOARD:

Here,
We can see that number of siblings/spouses=1 has highest survival rate.

# FINDING NUMBER OF SIBLINGS/SPOUSES ABOARD:

# FINDING NUMBER OF PARENTS/CHILDREN ABOARD:

33 | P a g e
# FINDING AVERAGE SURVIVAL RATE WITH RESPECT TO
NUMBER OF PARENTS/CHILDREN ABOARD:

Here,
We can see that number of parents/children=3 has highest survival rate.

# FINDING FIRST TEN TICKET NUMBER:

# FINDING NUMBER OF TICKETS FOR DIFFERENT TICKET

NUMBER LENGTH:

34 | P a g e
# FINDING AVERAGE SURVIVAL RATE WITH RESPECT TO TICKET
NUMER LENGTH:

# FINDING NUMBER OF PEOPLE FOR DIFFERENT FARE GROUP:

Here,
We can see that Fare Group 0 to 8.662 has highest number of people.

# FINDING AVERAGE SURVIVAL RATE FOR DIFFERENT FARE

GROUP:

Here,
We can see that Fare Group 26.0 to 512.329 has highest number of people survived.

35 | P a g e
# FINDING CROSSTAB

The crosstab method is used to compute a simple cross-tabulation of two (or more) factors. By
default, computes a frequency table of the factors unless an array of values and an aggregation
function are passed. Here cross-tabulation is done between ‘Fare’ and ‘Pclass’. ‘Fare’ is divided
into some groups and cross-tabulation is done for three types of ‘Pclass’.

# FINDING FIRST LETTER OF DIFFERENT CABINS AND NUMBER OF

PEOPLE FOR DIFFERENT CABINS:

Here,
We can see the first letter of cabin ‘n’ has highest number of people.

36 | P a g e
# FINDING AVERAGE SURVIVAL RATE OF PASSENGERS WITH
RESPECT TO FIRST LETTER OF DIFFERENT CABINS:

# FINDING NUMBER OF PEOPLE IN DIFFERENT CABIN NUMBERS:

# FINDING AVERAGE SURVIVAL RATE FOR DIFFERENT CABIN

NUMBER:

37 | P a g e
# FINDING CORRELATION BETWEEN CABIN NUMBER AND
SURVIVED:

It shows negative correlation, with increasing cabin number survival rate is decreasing.

# FINDING NUMBER OF PASSENGERS FOR DIFFERENT PORT OF

EMBARKATION

Here,
We can see that Port of Embarkation ‘S’ has highest number of passengers.

# FINDING AVERAGE SURVIVAL RATE WITH RESPECT TO

DIFFERENT PORT OF EMBARKATION:

38 | P a g e
# PLOTTING COUNTPLOT OF DIFFERENT PORT OF EMBARKATION
WITH RESPECT TO ‘PCLASS’:

Here,
We can see that Port of Embarkation ‘S’ has highest number of passengers in every Pclass. It
has highest people in Pclass = 3.

# DIFFERENT FUNCTIONS AND USAGE:

This function creates two separate columns: a numeric column indicating the length of a
passenger's Name field, and a categorical column that extracts the passenger's title. looking at
the relationship between the length of a name and survival rate appears to indicate that there is
indeed a clear relationship.

39 | P a g e
In this function the null values of the Age column are imputed by filling with the mean value of
the passengers’ age corresponding to title and class. This more granular approach to imputation
should be more accurate than merely taking the mean age of the population.

In this function we are creating a new variable called ‘Fam_Size’ for family size. The
importance of this variable is high because the distribution and survival rate between the
different categories does not give much hope. Same conclusions as Sibsp: passengers with zero
parents or children had a lower likelihood of survival than otherwise, but that survival rate was
only slightly less than the overall population survival rate. We combine the SibSp and Parch
columns into a new variable that indicates family size, and group the family size variable into
three categories.

Another piece of information is the first letter of each ticket, which, again, might be indicative
of a certain attribute of the ticketholders or their rooms. The Ticket column is used to create two
new columns: Ticket_Lett, which indicates the first letter of each ticket (with the smaller-n
values being grouped based on survival rate); and Ticket_Len, which indicates the length of the
Ticket field.

40 | P a g e
We can see that most of the cabin letters are associated with a high survival rate, so this might
very well be a useful variable. Because there aren't that many unique values, we won't do any
grouping here, even if some of the values have a small count.

Upon first glance, this appears to be useless. Not only do we have ~700 nulls which will be
difficult to impute, but the correlation with Survived is almost zero. However, the cabin
numbers as a whole do seem to have a high survival rate compared to the population average,
so we can use it as a feature.

In this function we fill the null values in the Embarked column with the most commonly
occurring value, which is 'S.'

41 | P a g e
because we are using scikit-learn, we must convert our categorical columns into dummy
variables. The following function does this, and then it drops the original categorical columns.
It also makes sure that each category is present in both the training and test datasets.

Our last helper function drops any columns that haven't already been dropped. In our case, we
only need to drop the PassengerId column, which we have decided is not useful for our problem
(by the way, I've confirmed this with a separate test). Note that dropping the PassengerId
column here means that we'll have to load it later when creating our submission file.

We also fill in the one missing value of Fare in our test set with the mean value of Fare from the
training set (transformations of test set data must always be fit using training data).

42 | P a g e
# BUILDING THE MODEL:

Having built our helper functions, we can now execute them in order to build our dataset that
will be used in the model.

We can see that our final dataset has 45 columns, composed of our target column and 44
predictor variables. Although highly dimensional datasets can result in high variance, I think we
should be fine here.

43 | P a g e
5.3 HYPERPARAMETER TUNING

We will use grid search to identify the optimal parameters of our random forest model. Because
our training dataset is quite small, we can get away with testing a wider range of
hyperparameter values.

Looking at the results of the grid search:

0.8293001521701072
{'criterion': 'gini', 'min_samples_leaf': 5, 'min_samples_split': 12, 'n_estimators': 700}

...we can see that our optimal parameter settings are not at the endpoints of our provided values,
meaning that we do not have to test more values. What else can we say about our optimal
values? The min_samples_split parameter is at 12, which should help mitigate overfitting to a
certain degree. This is especially good because we have a relatively large number of estimators
(700), which could potentially increase our generalization error.

44 | P a g e
5.4 MODEL ESTIMATION AND EVALUATION:

Random Forest Classifier:

The random forest classifier is a supervised learning algorithm which you can use for regression
and classification problems. It is among the most popular machine learning algorithms due to its
high flexibility and ease of implementation.

Why is the random forest classifier called the random forest?

That’s because it consists of multiple decision trees just as a forest has many trees. On top of
that, it uses randomness to enhance its accuracy and combat overfitting, which can be a huge
issue for such a sophisticated algorithm. These algorithms make decision trees based on a
random selection of data samples and get predictions from every tree. After that, they select the
best viable solution through votes.

It has numerous applications in our daily lives such as feature selectors, recommender systems,
and image classifiers. Some of its real-life applications include fraud detection, classification of
loan applications, and disease prediction. It forms the basis for the Boruta algorithm, which
picks vital features in a dataset.

We are now ready to fit our model using the optimal hyperparameters. The out-of-bag score can
give us an unbiased estimate of the model accuracy, and we can see that the score is 82.63%.
Let's take a brief look at our variable importance according to our random forest model. We can
see that some of the original columns we predicted would be important in fact were, including
45 | P a g e
gender, fare, and age. But we also see title, name length, and ticket length feature prominently,
which are also used as useful variables.

Our last step is to predict the target variable. Here we have got the array of prediction of survival.
Here, 1 = Survived, 0 = Not Survived.

46 | P a g e
5.5 CONFUSION MATRIX:

Confusion matrix is a very popular measure used while solving classification problems. It can
be applied to binary classification as well as for multiclass classification problems. An example
of a confusion matrix for binary classification is shown in Table .

Table . Confusion matrix for binary classification.

Predicted
Actual Negative Positive
Negative TN FP
Positive FN TP

Confusion matrices represent counts from predicted and actual values. The output “TN” stands
for True Negative which shows the number of negative examples classified accurately.
Similarly, “TP” stands for True Positive which indicates the number of positive examples
classified accurately. The term “FP” shows False Positive value, i.e., the number of actual
negative examples classified as positive; and “FN” means a False Negative value which is the
number of actual positive examples classified as negative. One of the most commonly used
metrics while performing classification is accuracy.
The accuracy of a model (through a confusion matrix) is calculated using the given formula
below.

Accuracy can be misleading if used with imbalanced datasets, and therefore there are other
metrics based on confusion matrix which can be useful for evaluating performance.

47 | P a g e
Here we can see,

TN = 123,

TP = 65,

FN = 21,

FP =14

Here ,

True Negative (TN) means passenger did not survive and it is predicted by the model correctly.

True Positive(TP) means passenger survived and it is predicted by the model correctly.

False Negative (FN) means passenger survived but it is predicted by the model incorrectly that
the passenger did not survive.

False Positive(FP) means passenger did not survive but it is predicted by the model incorrectly
that the passenger survived.

So, we are getting accuracy score = 0.843

48 | P a g e
# PLOTTING NORMALIZED CONFUSION MATRIX:

The higher the diagonal values of the confusion matrix the better, indicating many correct
predictions. A normalized confusion matrix makes it easier for the data scientist to visually
interpret how the labels are being predicted.

TN = 0.89

FN = 0.27

FP = 0.11

TP = 0.73

49 | P a g e
5.6 CLASSIFICATION REPORT:

A Classification report is used to measure the quality of predictions from a classification

algorithm. How many predictions are True and how many are False. More specifically, True
Positives, False Positives, True negatives and False Negatives are used to predict the metrics of
a classification report as shown below.

The report shows the main classification metrics precision, recall and f1-score on a per-class
basis. The metrics are calculated by using true and false positives, true and false negatives.
Positive and negative in this case are generic names for the predicted classes. There are four
ways to check if the predictions are right or wrong:

TN / True Negative: when a case was negative and predicted negative

TP / True Positive: when a case was positive and predicted positive

FN / False Negative: when a case was positive but predicted negative

FP / False Positive: when a case was negative but predicted positive

Precision – What percent of your predictions were correct?

Precision is the ability of a classifier not to label an instance positive that is actually negative.
For each class it is defined as the ratio of true positives to the sum of true and false positives.

TP – True Positives
FP – False Positives

Precision – Accuracy of positive predictions.

Precision = TP/(TP + FP)

50 | P a g e
Recall – What percent of the positive cases did you catch?

Recall is the ability of a classifier to find all positive instances. For each class it is defined as
the ratio of true positives to the sum of true positives and false negatives.

FN – False Negatives

Recall: Fraction of positives that were correctly identified.

Recall = TP/(TP+FN)

F1 score – What percent of positive predictions were correct?

The F1 score is a weighted harmonic mean of precision and recall such that the best score is 1.0
and the worst is 0.0. Generally speaking, F1 scores are lower than accuracy measures as they
embed precision and recall into their computation. As a rule of thumb, the weighted average of
F1 should be used to compare classifier models, not global accuracy.

F1 Score = 2*(Recall * Precision) / (Recall + Precision)

Support

Support is the number of actual occurrences of the class in the specified dataset. Imbalanced
support in the training data may indicate structural weaknesses in the reported scores of the
classifier and could indicate the need for stratified sampling or rebalancing. Support doesn’t
change between models but instead diagnoses the evaluation process.

Here,

For ‘Not-Survived’- Precision: 0.85 , Recall: 0.90, F1-Score: 0.88, Support: 137

And for ‘Survived’- Precision: 0.82 , Recall: 0.76, F1-Score: 0.79, Support: 86

The accuracy is : 0.84

51 | P a g e
5.7 LEARNING CURVE:

A learning curve is a plot of model learning performance over experience or time.

Learning curves are a widely used diagnostic tool in machine learning for algorithms that learn
from a training dataset incrementally. The model can be evaluated on the training dataset and on
a hold out validation dataset after each update during training and plots of the measured
performance can created to show learning curves.

Reviewing learning curves of models during training can be used to diagnose problems with
learning, such as an underfit or overfit model, as well as whether the training and validation
datasets are suitably representative.

Underfit Learning Curves

Underfitting refers to a model that cannot learn the training dataset. An underfit model can be
identified from the learning curve of the training loss only.

It may show a flat line or noisy values of relatively high loss, indicating that the model was
unable to learn the training dataset at all.

An underfit model may also be identified by a training loss that is decreasing and continues to
decrease at the end of the plot.

This indicates that the model is capable of further learning and possible further improvements
and that the training process was halted prematurely.

A plot of learning curves shows underfitting if:

• The training loss remains flat regardless of training.

• The training loss continues to decrease until the end of training.

Overfit Learning Curves

Overfitting refers to a model that has learned the training dataset too well, including the
statistical noise or random fluctuations in the training dataset. The problem with overfitting, is
that the more specialized the model becomes to training data, the less well it is able to
generalize to new data, resulting in an increase in generalization error. This increase in
generalization error can be measured by the performance of the model on the validation dataset.

This often occurs if the model has more capacity than is required for the problem, and, in turn,
too much flexibility. It can also occur if the model is trained for too long.

A plot of learning curves shows overfitting if:

• The plot of training loss continues to decrease with experience.

• The plot of validation loss decreases to a point and begins increasing again.

52 | P a g e
The inflection point in validation loss may be the point at which training could be halted as
experience after that point shows the dynamics of overfitting.

Good Fit Learning Curves

A good fit is the goal of the learning algorithm and exists between an overfit and underfit
model.

A good fit is identified by a training and validation loss that decreases to a point of stability
with a minimal gap between the two final loss values.

The loss of the model will almost always be lower on the training dataset than the validation
dataset. This means that we should expect some gap between the train and validation loss
learning curves. This gap is referred to as the “generalization gap.”

A plot of learning curves shows a good fit if:

• The plot of training loss decreases to a point of stability.

• The plot of validation loss decreases to a point of stability and has a small gap with the
training loss.
Continued training of a good fit will likely lead to an overfit.

From the curve, we can clearly see that as the size of the training set increases, distance
between the training score curve and the cross-validation score curve decreases. The cross-
validation accuracy increases as we add more training data. So adding training data is useful in
this case. Since the training score is very accurate, this indicates low bias and high variance. So

53 | P a g e
this model also begins overfitting the data because the cross-validation score is relatively lower
and increases very slowly as the size of the training set increases.

5.8 PREDICTIONS FOR TEST DATASET:

Here we have got the array of prediction of survival.

Here, 1 = Survived, 0 = Not Survived.

54 | P a g e
# WRITING FINAL PREDICTION OF TEST DATASET TO A
SEPARATE CSV FILE:

55 | P a g e
 CHAPTER-6
FINDINGS & DISCUSSION
• We can see that number of survived is less than number of not survived.

• From this plot , we can see that number of male passengers is greater than that of
female.

• Pclass = 3 has maximum number of passengers.

• In case of Port of Embarkation(C = Cherbourg; Q = Queenstown; S = Southampton), S

has highest number of passengers.

• Here, We can see that female has higher survival rate than that of male.

• We find that ‘Lady’,’Mlle’, ‘Mme’, ‘ Ms’, ‘Sir’, ‘the’ – these titles have highest survival
rate.

• Here, we can see that Age group 0.419 to 19.0 has highest survival rate. Age group 19.0
to 25.0 has lowest survival rate.

• Age group 0.419 to 19.0 has highest number of people. Age group 25.0 to 31.0 has
lowest number of people.

• Number of siblings/spouses=1 has highest survival rate.

• Number of parents/children=3 has highest survival rate.

• Here, we can see that Fare Group 0 to 8.662 has highest number of people.

• Fare Group 26.0 to 512.329 has highest number of people survived.

• The first letter of cabin ‘n’ has highest number of people.

• It shows negative correlation , with increasing cabin number survival rate is decreasing.

• Port of Embarkation ‘S’ has highest number of passengers in every Pclass. It has highest
people in Pclass = 3.

• Sibsp: passengers with zero parents or children had a lower likelihood of survival than
otherwise, but that survival rate was only slightly less than the overall population
survival rate.

• From Model Estimation and Evaluation we get the out-of-bag score.

The out-of-bag score can give us an unbiased estimate of the model accuracy, and we
56 | P a g e
 can see that the score is 82.63%.

• From Confusion Matrix we get,

TN = 123,

TP = 65,

FN = 21,

FP =14
So, we are getting accuracy score = 0.843

• From Classification Report we get:

For ‘Not-Survived’- Precision: 0.85 , Recall: 0.90, F1-Score: 0.88, Support: 137

And for ‘Survived’- Precision: 0.82 , Recall: 0.76, F1-Score: 0.79, Support: 86

The accuracy is : 0.84

• From the Learning curve, we can clearly see that as the size of the training set increases,
distance between the training score curve and the cross-validation score curve decreases.
The cross-validation accuracy increases as we add more training data. So adding
training data is useful in this case. Since the training score is very accurate, this indicates
low bias and high variance. So this model also begins overfitting the data because the
cross-validation score is relatively lower and increases very slowly as the size of the
training set increases.

57 | P a g e
 CHAPTER-7
IMPLEMENTATION

The implementation of this project involved the following steps: -

• Import the necessary packages

• Import the required dataset and clean the dataset

• Visualize the dataset

• Split the dataset into a training and test set

• Apply the algorithm

• Generate graphs comparing train and test accuracies

• Generate Confusion Matrix, Classification Report and Learning curve

58 | P a g e
 CHAPTER-8
CONCLUSIONS

We can conclude that ,

• We are getting accuracy score = 0.843, this means our model is not 100% accurate.

• From Classification Report we get:

For ‘Not-Survived’- Precision: 0.85 , Recall: 0.90, F1-Score: 0.88, Support: 137

And for ‘Survived’- Precision: 0.82 , Recall: 0.76, F1-Score: 0.79, Support: 86

which shows that our algorithm is not that accurate, but it can still make good
predictions.

In this story, we applied the concepts of Random Forest on Titanic Dataset.

59 | P a g e
 CHAPTER-9
REFERENCES

❖ https://towardsdatascience.com/predicting-the-survival-of-titanic-passengers-30870ccc7e8

❖ https://www.kaggle.com/zlatankr/titanic-random-forest-82-78

❖ https://www.datacourses.com/evaluation-of-regression-models-in-scikit-learn-846/

❖ https://realpython.com/best-python-books/

❖ https://machinelearningmastery.com/machine-learning-in-python-step-by-step/

❖ https://www.w3schools.com/python/python_ml_getting_started.asp

60 | P a g e
 THANK YOU

61 | P a g e