SKIN DISEASE PREDICTION AND CLASSIFICATION

ANALYSIS – A DEEP LEARNING APPROACH

A PROJECT REPORT

Submitted in partial fulfillment of the requirements for the award of the degree of

Bachelor of Technology

in

COMPUTER SCIENCE AND ENGINEERING

BY
Student-1 Name Student-2 Name
(Roll Number) (Roll Number)
Student-3 Name Student-4 Name
(Roll Number) (Roll Number)
Student-5 Name
(Roll Number)
Under the Supervision of
Dr. P. Rama Santosh Naidu
Senior Assistant Professor

DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING

MAHARAJ VIJAYARAM GAJAPATHI RAJ COLLEGE OF ENGINEERING
(Autonomous)
(Approved by AICTE, New Delhi, and permanently affiliated to JNTUGV, Vizianagaram), Listed u/s 2(f)
& 12(B) of UGC Act 1956.
Vijayaram Nagar Campus, Chintalavalasa,Vizianagaram-535005, Andhra Pradesh
APRIL, 2024
 CERTIFICATE

This is to certify that the project report entitled “SKIN DISEASE PREDICTION AND
CLASSIFICATION ANALYSIS – A DEEP LEARNING APPROACH” being submitted
by Ch.Spandana(18331A0527), B.Adarsh(18331A0518), B.Srinidhi(18331A0511),
B.Dilleswara Rao(19335A0502), B.Prudhvi Raj(19335A0503) in partial fulfillment for the
award of the degree of “Bachelor of Technology” in Computer Science and Engineering
is a record of bonafide work done by them under my supervision during the academic year
2021-2022.

Dr. P. Rama Santosh Naidu Dr. T. Pavan Kumar

Senior Assistant professor, Associate Professor,
Supervisor, Head of the Department,
Department of CSE, Department of CSE,
MVGR College of Engineering(A), MVGR College of Engineering(A),
Vizianagaram. Vizianagaram.

External Examiner
 DECLARATION

We hereby declare that the work done on the dissertation entitled “SKIN DISEASE
PREDICTION AND CLASSIFICATION ANALYSIS – A DEEP LEARNING
APPROACH” has been carried out by us and submitted in partial fulfilment for the award of
credits in Bachelor of Technology in Computer Science and Engineering of MVGR College
of Engineering (Autonomous) and affiliated to JNTUGV, Vizianagaram. The various
contents incorporated in the dissertation have not been submitted for the award of any degree
of any other institution or university.
 ABSTRACT
Skin diseases are very common for humans in day-to-day lives. The task of recognizing the
type of skin disease and prediction of its occurrence is the most complicated one as we have
hundreds of varieties of skin textures and different visual closeness effects of diseases and
other factors. Therefore, it is of high importance to detect the occurrence of any skin disease
and also, we should be able to find the type of skin disease to get best treatment.
The proposed system focuses on building the model for the prediction and classification
analysis of different types of skin diseases using deep learning techniques and algorithms.
These deep learning algorithms help the model in predicting and classifying the skin diseases
on different characteristics of the dataset with more accuracy.
 CONTENTS
Page No
List of Abbreviations 1
List of Tables 2
List of Figures 3
1. Introduction 4
1.1. Identification of seriousness of the problem 5
1.2. Problem definition 5
1.3. Objective 6
1.4. Existing models 6
2. Literature Survey 7
3. Theoretical Background 10
3.1. Machine learning Vs Deep Learning 10
3.1.1. What is Machine Learning 10
3.1.2. Why Machine Learning 10
3.1.3. What is Deep Learning 10
3.1.4. Why Deep Learning 11
3.2. Machine Learning Approaches 11
3.2.1 Supervised Learning 11
3.2.2 Unsupervised Learning 11
3.2.3 Semi-Supervised Learning 12
3.2.4 Reinforcement Learning 12
3.3 Machine Learning Models 12
3.3.1 Artificial Neural Networks 12
3.3.2 Support Vector Machine 12
3.3.2.1 Linear SVM 13
3.3.2.2 Non-Linear SVM 13
3.3.3 Logistic Regression 13
3.3.3.1 Types of Logistic Regression 13
3.3.4 K-nearest neighbor Algorithm 14
3.3.5 Naïve Bayes 14
3.4 Confusion matrix and its metrics 15
3.4.1 Confusion matrix 15
3.4.2 AUC-ROC curve 16
3.4.3 Metrics- Accuracy, Precision, Recall, F1 Score, FPR 16
3.5 Exploratory Data Analysis 17
4. Approach Description 18
4.1. Approach Flow 18
5. Data Exploration 20
5.1. Dermatology dataset 20
5.1.1. Data Manipulation 20
5.1.2. Data Preparation 20
5.1.2.1. Missing Values 21
 5.1.2.2. Duplicate values 21
5.1.2.3. Outliers 21
6. Data Analysis 23
6.1. EDA and its types 23
6.2. Why do we need data analysis? 23
6.3. EDA Inferences 23
6.3.1. Skin Disease Vs Count 23
6.3.2. Age Vs Disease 24
6.3.3. Itching factor Vs Disease 24
6.3.4. Family Vs Count 25
6.3.4.1. Family history-yes Vs Count 25
6.3.4.2. Family history-no Vs Count 26
6.4. Correlation Heatmap 26
7. Modelling 28
7.1. Model Development 28
7.1.1. Multi-class classification problem 28
7.1.2. Transformation to binary 28
7.1.3. Extension from binary 29
7.1.3.1. Naïve Bayes classifier 29
7.2. Model Evaluation 30
8. Results and Conclusions 31
References 32
Appendix A: Packages, Tools Used & Working Process 35
Python Programming Language 35
Libraries 35
NumPy 35
Pandas 36
Matplotlib 36
Seaborn 36
Sklearn 37
Tools used Jupyter Notebooks 38
Appendix B: Source Code 40
 List of Abbreviations
AI – Artificial Intelligence
ML – Machine Learning
NN – Neural Networks
ANN – Artificial Neural Networks
CNN – Convolutional Neural Networks
SVM – Support Vector Machine
KNN – K-nearest neighbors
RGB – Red Green Blue
EDA – Exploratory Data Analysis
IDA – Integrated Data Analysis
CSV – Comma Separated Values
DML – Data Manipulation Language
SIFT – Scale Invariant Feature Transform
OVR – One Vs Rest
OVO – One Vs One
PCA – Principal Component Analysis
AUC – Area under the Curve
ROC – Receiver Operating Characteristics curve
TP – True positive
TN – True Negative
FP – False Positive
FN – False Negative
TPR – True Positive Rate
FPR – False Positive Rate

pg. 1
 List of Figures
Figure 1.1 : Bar diagram showing distribution of sex according to the age
Figure 3.1 : Confusion matrix with 4 parameters
Figure 5.1 : Checking for number of missing values in each column
Figure 5.2 : Checking for number of duplicate rows in dataset
Figure 5.3 : Check and remove outliers in age column with histogram
Figure 5.4 : Check and remove outliers like special chars in age column
Figure 6.1 : Counterplot for Skin disease Vs Count
Figure 6.2 : Distribution of Age according to the Skin disease
Figure 6.3 : Itching factor Vs Skin disease
Figure 6.4 : Family history Vs Count
Figure 6.5 : Family history-yes Vs Count
Figure 6.6 : Family history-no Vs Count
Figure 6.7 : Correlation Heatmap

pg. 2
 List of Tables
Table 1.1 : Sex distribution according to the age of the surveyed population
Table 8.1 : Results of Performance metrics of each class label

pg. 3
 CHAPTER 1
INTRODUCTION
Dermatological problems are the major widespread diseases in the world that are very
common in our daily lives. The diagnosis of these skin diseases is highly complicated and is
not accurate due to various kinds and complexities in skin complexions, tone, and presence of
hair on the skin, etc., So we identify dermatology as one of the most complicated branches of
science [1]. We have these variations and varieties in skin diseases and skin types due to
many changes made in the environment, geographical changes, biological changes, pollution,
etc., These skin diseases can spread from one person to another. So, it is important to treat
them at the early stages to avoid problems. The total well-being of a person both physically
and mentally is also affected due to these skin diseases in a wide range. There might be a
threat to life sometimes due to the severity of the disease. The prevalence of skin diseases has
increased over the past few decades and they contribute to a significant burden on healthcare
systems across the world. So, there is a need to have an instrument or model to detect and
identify the type of skin disease at the earliest stage possible and give proper treatment for
each type of skin disease without facing unexpected life risk issues.
According to various studies and surveys, the existence of skin diseases among the general
population varied from 8% to 11%. In the year 2017, the number of age-standardized years
in India who lived with disability for cardio-vascular diseases was nearly 333 and for skin
and subcutaneous diseases was nearly 455 per 100,000. As per the global burden of disease
study in 2017, cardiovascular diseases ranked 12 th whereas skin diseases ranked 10th based on
these age-standardized years lived with disability.

35

30

25

20

15

10

0
11 to 20 21 to 30 31 to 40 41 to 50 51 to 60 above 60
Female Male

Figure 1.1: Bar diagram showing distribution of sex according to the age

pg. 4
 Table 1.1: Sex distribution according to the age of the surveyed population

Age in years Female Male Total

11-20 8(2.3) 15(5.4) 23(3.7)
21-30 85(24.5) 82(29.7) 167(26.8)
31-40 85(24.5) 54(19.6) 139(22.3)
41-50 58(16.7) 31(11.2) 89(14.3)
51-60 46(13.3) 32(11.6) 78(12.5)
60+ 65(18.7) 62(22.5) 127(20.4)
Total 347(55.7) 276(44.3) 623(100.0)

Table 1.1 and Figure 1.1 shows the survey results of people based on their age and sex for
ages ranging from 11 and above. From the results of the survey, we can say that there are
55.7% females and 44.3% males. Almost half of the people i.e., 52.8% are of the age less
than 41 years and the remaining people of age 60 years and above constitute 20.8%.
1.1 Identification of seriousness of the problem
The diagnosis of these skin diseases is highly complicated due to various kinds and
complexities in skin complexions, tone, and presence of hair on the skin, etc., These skin
diseases are infectious and need to be treated at earlier stages to avoid spreading from one
person to other. The total well-being of a person both physically and mentally is also affected
due to these skin diseases in a wide range. There might be a threat to life sometimes due to
the severity of the disease. So, there is a need to have an instrument or model to detect and
identify the type of skin disease at the earliest stage possible and give proper treatment for
each type of skin disease without facing unexpected life risk issues.
According to various studies and surveys, the existence of skin diseases among the general
population varied from 8% to 11%. In the year 2017, the number of age-standardized years
in India who lived with disability for cardio-vascular diseases was nearly 333 and for skin
and subcutaneous diseases was nearly 455 per 100,000. As per the global burden of disease
study in 2017, cardiovascular diseases ranked 12 th whereas skin diseases ranked 10th based on
these age-standardized years lived with disability.
1.2 Problem definition
The data is collected from various sources and surveys to find patterns in the attributes and
factors causing skin disease. The important factors like age, family history, itching factor, and
some other factors are focused mainly in the data. The primary concern is to analyze these

pg. 5
important factors in order to predict the occurrence of a skin disease and the type of skin
disease occurred.
1.3 Objective
With the help of the model, we construct for identifying the skin diseases in humans, we
intend to find out the patterns in the data gathered by analyzing different factors which helps
to achieve our goal of prediction and classification of the skin disease that occurred so that
we can treat it at the earliest possible stage and avoid major life risks.
1.4 Existing models
There are many researchers and models previously and some are still going on to identify
skin diseases in different regions of a human being. But it doesn’t focus on certain issues to
be analyzed like family history, intensity of disease etc., and also prediction and classification
with good accuracy is also not so good.

pg. 6
 CHAPTER 2
LITERATURE SURVEY
Skin diseases are hazardous to life. They should be detected at an early stage to cure them
properly without spreading. But the task of detection is very complicated. Several researchers
proposed various methodologies and technologies for accurate and appropriate detection of
skin diseases.
In [2], some researchers suggested skin disease detection on the basis of image processing.
Here they used Red Green Blue spectrum (RGB) images of the areas affected by skin
diseases as input. These images are re-sized and features are extracted using a pre-trained
CNN (Convolution Neural Network) model and then they applied multi support vector
machine (SVM) to classify. Finally, they have shown that this is the simplest method with
100% accuracy. In [3], researchers have suggested a model using SVM and KNN algorithms
for the purpose of segmentation and classification of a skin disease occurred. In other paper
[4], Researchers used an artificial intelligence (AI) system on the basis of neural network
(NN) concepts. This system is divided into two parts. They are feature extraction and
classification. Feature extraction is done by image acquisition and classification is done by
feed-forward neural network.
Other researchers in paper [5] studied six different algorithms for the classification of
different skin diseases. They chose 15 most important features for prediction of the disease
class to which it belongs. In addition to these 6 algorithms, some authors also created an
ensemble method using the combination of other classifier techniques too. At last, keeping all
the observations in mind, they concluded that more accurate and very effective skin disease
detection and prediction is done using the ensemble methods.
In [6] a researcher used three techniques – Segmentation which is done by thresholding,
Extraction of features using two-dimensional (2-D) wavelet decomposition and then
classification performed using radial basic neural network and back propagation neural
network to achieve the results. Researchers in [7] have proposed a model on the basis of
adaptive federated machine learning technique for skin diseases detection. This process has
an architecture of intelligent local edges and a central (global point) known as a server to
diagnosis the type of skin, skin diseases and also to improve the accuracy.

pg. 7
In [8] researchers have introduced a model using the methods of extracting features and
classification of images for detection of six different types of skin diseases and their
conditions.
In [9], other researcher has proposed a system using ad boost classifier and statistical analysis
for identification of skin problem at an early stage. Their research focused mainly on
identifying the symptoms of skin cancer at an early stage on the basis of statistical analysis
obtained by using correlation algorithms. In [10], researcher has proposed a model on the
basis of feature extraction of skin image using color texture and to identify the skin disease
they used segmentation and SVM. In [11], researcher has proposed a system based on
computer vision technique for skin disease detection which could be implemented on both
mobile and computer using desktop applications.
In another research paper [12], the researchers used five machine learning algorithms
Logistic Regression, random forest, Naïve Bayes, kernel SVM, and Convolution Neural
Network algorithms for skin diseases detection. Finally, from the confusion matrix, they
found that the convolution neural network (CNN) model is the best model for skin disease
detection. In [13], some other researchers have proposed the usage of various kinds of image
processing algorithms for feature extraction and feed forwarding using artificial neural
network to train and test the model. This work is of two parts, feature extraction and
classification. The feature extraction takes place on the basis of color texture and the
classifier finds the possible skin disease. In [14] researchers have proposed a model to
identify the psoriasis using color feature extraction and classification of the skin disease. In
this paper, Abbadi has mentioned color feature extraction method where color features are
extracted using a mathematical formula used for RGB color value and texture extraction
method where texture features are extracted using various components such as contract,
energy, entropy, and homogeneity of the skin image. After that Neural Network (NN)
algorithms are used to detect the psoriasis present on the skin. In [15] researchers have
proposed a model using machine learning and computer vision concepts. The feature
extraction of images is done and various algorithms are applied on these features for the
detection of six types of skin diseases with an accuracy of 95%. In [16] a researcher has
suggested an image clustering method using naïve bayes classifier. They have made use of
Scale Invariant Feature Transform (SIFT) method to detect key points of image. Later they
used SVM and CNN for the tasks of segmentation and classification with 84% accuracy and
82% precision. In [17], the authors proposed an artificial intelligence (AI) system consisting
of image processing and a deep neural network for the detection of skin cancer. First, they

pg. 8
performed segmentation on the affected area and extracted features using image processing,
and used a convolution neural network for prediction. This resulted in 93.7% accuracy for
training dataset and 89.5% accuracy for testing dataset. In [18], researchers have proposed a
two-stage method for skin disease prediction on the basis of color texture-based identification
and used a classification technique to identify the disease name. The first stage has 95.99%
accuracy and the second stage has a little less accuracy compared to the first stage i.e.,
94.016%. In [19], a researcher proposed that features are extracted with the help of graph cut
algorithms which are used for image processing of skin images. For classification, he used
naïve Bayes algorithm as a classification algorithm.

pg. 9
 CHAPTER 3
THEORETICAL BACKGROUND
3.1 Machine learning Vs Deep learning
3.1.1 What is Machine Learning?
Machine learning is an application of AI that enables systems to learn and improve from
experience without being explicitly programmed. Machine learning focuses on developing
computer programs that can access data and use it to learn for themselves. Machine learning
is something that is capable to imitate the intelligence of the human behavior. Machine
learning is used to perform complex tasks in a way that humans solve the problems. Machine
learning can be descriptive it uses the data to explain, predictive, and prescription.
3.1.2 Why Machine Learning?
Machine learning involves computers learning from data provided so that they carry out
certain tasks. For more advanced tasks, it can be challenging for a human to manually create
the needed algorithms. In practice, it can turn out to be more effective to help the machine
develop its own algorithm, rather than having human programmers specify every needed step.
The discipline of machine learning employs various approaches to teach computers to
accomplish tasks where no fully satisfactory algorithm is available. In cases where vast
numbers of potential answers exist, one approach is to label some of the correct answers as
valid. This can then be used as training data for the computer to improve the algorithms it
uses to determine correct answers. The nearly limitless quantity of available data,
affordable data storage, and growth of less expensive and more powerful processing has
propelled the growth of ML. Now many industries are developing more robust models
capable of analysing bigger and more complex data while delivering faster, more accurate
results on vast scales. ML tools enable organizations to more quickly identify profitable
opportunities and potential risks.
The practical applications of machine learning drive business results which can
dramatically affect a company’s bottom line. New techniques in the field are evolving
rapidly and expanded the application of ML to nearly limitless possibilities. Industries that
depend on vast quantities of data—and need a system to analyse it efficiently and
accurately, have embraced ML as the best way to build models, strategize, and plan.

pg. 10
3.1.3 What is Deep Learning?
Deep learning is the subset of the machine learning technique. It is also known as deep neural
network because it uses the architecture of neural network. Deep learning eliminates some of
data preprocessing. This has one input layer and one output layer and more than one hidden
layer. This uses labeled data for training deep learning prediction is more accurate. Deep
learning is an element of data science, it includes statistics and predictive modeling. This
approach is beneficial to the person whose task are related to collecting, analyzing large
amounts of data.
3.1.4 Why Deep Learning?
Algorithms used in deep learning learn high level features for data. The more the data you fed
to the deep learning the better the result will be. Deep learning handles large volumes of data
and it is kept to the best use when it comes to the large sets of unstructured data. Deep
learning has less accuracy when it is fed with the less data.
3.2 Machine Learning Approaches
Machine learning approaches are traditionally divided into three broad categories, depending
on the nature of the "signal" or "feedback" available to the learning system:
 Supervised learning
 Unsupervised learning
 Reinforcement learning
3.2.1 Supervised learning
Supervised learning is one of the machine learning approaches through which models are
trained using perfectly labelled training data and on the basis of that models predicts the
output.
Types of supervised learning algorithms include active learning, classification and regression.
Classification algorithms are used when the outputs are restricted to a limited set of values,
and regression algorithms are used when the outputs may have any numerical value within a
range. Similarity learning is an area of supervised machine learning closely related to
regression and classification, but the goal is to learn from examples using a similarity
function that measures how similar or related two objects are.
3.2.2 Unsupervised learning
Unsupervised learning algorithms take a set of data that contains only inputs, and find
structure in the data, like grouping or clustering of data points. The algorithms, therefore,
learn from test data that has not been labelled, classified or categorized. Instead of responding
to feedback, unsupervised learning algorithms identify commonalities in the data and react

pg. 11
based on the presence or absence of such commonalities in each new piece of data. A central
application of unsupervised learning is in the field of density estimation in statistics, such as
finding the probability density function. Though unsupervised learning encompasses other
domains involving summarizing and explaining data features.

3.2.3 Semi-supervised learning

Semi-supervised learning falls between unsupervised learning (without any labelled training
data) and supervised learning (with completely labelled training data). Some of the training
examples are missing training labels, yet many machine-learning researchers have found that
un-labelled data, when used in conjunction with a small amount of labelled data, can produce
a considerable improvement in learning accuracy. In weakly supervised learning, the training
labels are noisy, limited, or imprecise; however, these labels are often cheaper to obtain,
resulting in larger effective training sets.
3.2.4 Reinforcement learning
Reinforcement learning is feedback-based machine learning technique, and it aims to
maximize the rewards by their hit and trial actions. In reinforcement learning the model
learns automatically using feedbacks without any labeled data, unlike supervised learning and
since there is no labelled data, the model is used to learn from its experiences only.
Reinforcement learning is an area of machine learning concerned with how software agents
ought to take actions in an environment so as to maximize some notion of cumulative reward.
3.3 Machine Learning Models
Performing machine learning involves creating a model, which is trained on some training
data and then can process additional data to make predictions. Various types of models have
been used and researched for machine learning systems.
3.3.1 Artificial neural networks
Artificial neural networks (ANN) are a sub field of artificial intelligence which is simply
known as neural networks. Similar to the human brain that has neurons interconnected to one
another, artificial neural networks also have neurons that are interconnected to one another in
various layers of the networks. These neurons are known as nodes. ANN consists of 3 layers
they are input layer, output layer and hidden layer. Input layer accepts inputs in several
different formats provided by the programmer. The hidden layer presents in between input
and output layers. It performs all the operations to find hidden features and patterns. Output
layer provides the output based on all the calculations provided by the hidden layer.
3.3.2 Support Vector Machine

pg. 12
Support Vector Machine or SVM is one of the most popular Supervised Learning algorithms,
which is used for Classification as well as Regression problems. However, primarily, it is
used for Classification problems in Machine Learning. The goal of the SVM algorithm is to
create the best line or decision boundary that can segregate n-dimensional space into classes
so that we can easily put the new data point in the correct category in the future. This best
decision boundary is called a hyperplane. SVM chooses the extreme points/vectors that help
in creating the hyperplane. These extreme cases are called support vectors, and hence the
algorithm is termed as Support Vector Machine.
3.3.2.1 Linear SVM
Linear SVM is used for linearly separable data, which means if a dataset can be classified
into two classes by using a single straight line, then such data is termed as linearly separable
data, and the classifier is used as Linear SVM classifier.
3.3.2.2 Non-linear SVM
Non-Linear SVM is used for non-linearly separated data, which means if a dataset cannot be
classified by using a straight line, then such data is termed as non-linear data and the
classifier used is called Non-linear SVM classifier. In its most simple type, SVM doesn’t
support multiclass classification natively. It supports binary classification and separating data
points into two classes. For multiclass classification, the same principle is utilized after
breaking down the multi classification problem into multiple binary classification problems.
The idea is to map data points to high dimensional space to gain mutual linear separation
between every two classes. This is called a One-to-One approach, which breaks down the
multiclass problem into multiple binary classification problems. A binary classifier per each
pair of classes. Another approach one can use is One-to-Rest. In that approach, the
breakdown is set to a binary classifier per each class.
3.3.3 Logistic Regression
Logistic regression is a supervised learning classification algorithm used to predict the
probability of a target variable. The nature of target or dependent variable is dichotomous,
which means there would be only two possible classes. In simple words, the dependent
variable is binary in nature having data coded as either 1 (stands for success/yes) or 0 (stands
for failure/no). Mathematically, a logistic regression model predicts P(Y=1) as a function of
X. It is one of the simplest ML algorithms that can be used for various classification
problems.
3.3.3.1 Types of Logistic Regression

pg. 13
Based on the number of categories of target variables that can be predicted, Logistic
regression can be divided into following types –
 Binary or Binomial In such a kind of classification, a dependent variable will have
only two possible types either 1 and 0.
 Multinomial In such a kind of classification, dependent variables can have 3 or more
possible unordered types or the types having no quantitative significance.
 Ordinal In such a kind of classification, dependent variables can have 3 or more
possible ordered types or the types having a quantitative significance.
3.3.4 k-Nearest Neighbors Algorithm
The k-nearest neighbors’ algorithm (k-NN) is a non-parametric classification method and is
used for classification and regression. In both cases, the input consists of the k closest
training examples in the data set. The output depends on whether k-NN is used for
classification or regression: In k-NN classification, the output is a class membership. An
object is classified by a plurality vote of its neighbors, with the object being assigned to the
class most common among its k nearest neighbors’ (k is a positive integer, typically small). If
k = 1, then the object is simply assigned to the class of that single nearest neighbor. Both for
classification and regression, a useful technique can be to assign weights to the contributions
of the neighbors, so that the nearest neighbors contribute more to the average than the more
distant ones.
3.3.5 Naïve Bayes Classifier
Naïve Bayes algorithm is a supervised learning algorithm, which is based on Bayes
theorem and used for solving classification problems.
The Naïve Bayes algorithm is comprised of two words Naïve and Bayes, which can be
described as:
Naïve: It is called Naïve because it assumes that the occurrence of a certain feature is
independent of the occurrence of other features. Such as if the fruit is identified on the bases
of colour, shape, and taste, then red, spherical, and sweet fruit is recognized as an apple.
Hence each feature individually contributes to identify that it is an apple without depending
on each other.
Bayes: It is called Bayes because it depends on the principle of Bayes' Theorem
Bayes' Theorem:
Bayes' theorem is also known as Bayes' Rule or Bayes' law, which is used to determine the
probability of a hypothesis with prior knowledge. It depends on the conditional probability.
The formula for Bayes' theorem is given as:

pg. 14
where,
P(A|B) is Posterior probability: Probability of hypothesis A on the observed event B.
P(B|A) is Likelihood probability: Probability of the evidence given that the probability of a
hypothesis is true.
P(A) is Prior Probability: Probability of hypothesis before observing the evidence.
P(B) is Marginal Probability: Probability of Evidence.
Types of Naïve Bayes Model:
There are three types of Naive Bayes Model, which are given below:
 Gaussian: The Gaussian model assumes that features follow a normal distribution.
This means if predictors take continuous values instead of discrete, then the model
assumes that these values are sampled from the Gaussian distribution.
 Multinomial: The Multinomial Naïve Bayes classifier is used when the data is
multinomial distributed. It is primarily used for document classification problems, it
means a particular document belongs to which category such as Sports, Politics,
education, etc., The classifier uses the frequency of words for the predictors.
 Bernoulli: The Bernoulli classifier works similar to the Multinomial classifier, but the
predictor variables are the independent Booleans variables. Such as if a particular
word is present or not in a document. This model is also famous for document
classification tasks.
3.4 Confusion matrix and its metrics
3.4.1 Confusion matrix
A confusion matrix is a table that is often used to describe the performance of a classification
model on a set of test data for which the true values are known. All the measures can be
calculated by using left most four parameters. So, let’s talk about those four parameters first.

Figure 3.1: Confusion matrix with 4 parameters

pg. 15
True positive and true negatives are the observations that are correctly predicted. We want to
minimize false positives and false negatives. These terms are a bit confusing. So, let’s take
each term one by one and understand it fully.
True Positives (TP) - These are the correctly predicted positive values which means that the
value of actual class is yes and the value of predicted class is also yes.
True Negatives (TN) - These are the correctly predicted negative values which means that
the value of actual class is no and value of predicted class is also no.
False positives and false negatives, these values occur when your actual class contradicts with
the predicted class.
False Positives (FP) – When actual class is no and predicted class is yes.
False Negatives (FN) – When actual class is yes but predicted class in no.
3.4.2 AUC- ROC curve
The Receiver Operator Characteristic (ROC) curve is an evaluation metric for binary
classification problems. It is a probability curve that plots the TPR against FPR at various
threshold values and essentially separates the ‘signal’ from the ‘noise’. The Area Under
the Curve (AUC) is the measure of the ability of a classifier to distinguish between classes
and is used as a summary of the ROC curve.
The higher the AUC, the better the performance of the model at distinguishing between the
positive and negative classes.
When AUC = 1, then the classifier is able to perfectly distinguish between all the Positive
and the Negative class points correctly. If, however, the AUC had been 0, then the classifier
would be predicting all Negatives as Positives, and all Positives as Negatives. When
0.5<AUC<1, there is a high chance that the classifier will be able to distinguish the positive
class values from the negative class values. When AUC=0.5, then the classifier is not able to
distinguish between Positive and Negative class points.
3.4.3 Metrics
Accuracy
Accuracy is the most intuitive performance measure and it is simply a ratio of correctly
predicted observation to the total observations.
Accuracy = TP+TN/TP+FP+FN+TN
Precision
Precision is the ratio of correctly predicted positive observations to the total predicted
positive observations.
Precision = TP/TP+FP

pg. 16
Re-call/ Sensitivity/ True Positive Rate (TPR)
Recall is the ratio of correctly predicted positive observations to the all observations in actual
class - yes.
Recall = TP/TP+FN

F1 Score
F1 Score is the weighted average of Precision and Recall. Therefore, this score takes both
false positives and false negatives into account. Intuitively it is not as easy to understand as
accuracy, but F1 is usually more useful than accuracy, especially if you have an uneven class
distribution.
F1 Score = 2*(Recall * Precision) / (Recall + Precision)
False Positive Rate (FPR)
FPR tells us what proportion of the negative class got incorrectly classified by the classifier.
FPR=FP/TN+FP
3.5 Exploratory Data Analysis
Exploratory data analysis is an approach of analysing data sets to summarize their main
characteristics, often using statistical graphics and other data visualization methods. A
statistical model can be used or not, but primarily EDA is for seeing what the data can tell us
beyond the formal modelling or hypothesis testing task. EDA is different from initial data
analysis (IDA) which focuses more narrowly on checking assumptions required for model
fitting and hypothesis testing, and handling missing values and making transformations of
variables as needed. EDA encompasses IDA. Exploratory Data Analysis refers to the critical
process of performing initial investigations on data so as to discover patterns, to spot
anomalies, to test hypotheses and to check assumptions with the help of summary statistics
and graphical representations.

pg. 17
 CHAPTER 4
APPROACH DESCRIPTION
4.1. Approach Flow
1. We collected the data related to the dermatology by researching through the online
website [20].
2. Import the required packages for the classification model.
3. Creating a project and loading the dataset from the comma separated values (CSV)
file which is based on analysis of different variables.
4. Data cleaning:
 Check if there are any duplicate rows.
 Check the presence of any null values in each column.
 Overview the dermatology dataset (optional)
 Rename the 35 columns with the following names in sequential order:
'erythema','scaling','definite_borders','itching','koebner_phenomenon',
'polygonal_paples','follicular_papules','oral_mucosal_involvement','knee_elbo
w_involvement','scalp_involvement','family_history', 'melanin', 'eosinophils',
'PNL', 'fibrosis', 'exocytosis', 'acanthosis', 'hyperkeratosis', 'parakeratosis',
'clubbing', 'elongation', 'suprapapillary_epidermis', 'spongiform_pustule',
'munro_microabcess', 'focal_hypergranulosis',' granular_layer', 'basal_layer',
'spongiosis’,’sawtooth', 'follicular', 'perifollicular',
'inflammatory_monoluclear', 'band-like infiltrate', 'Age', 'target'
1. EDA - Exploratory data analysis (EDA) is a method to analyze the data sets and
acquire the summary of important attributes and their relationships through various
forms and techniques of data visualization based on some statistical data calculated.
a. Count of Different Target Values (Skin Disease Vs Count)

pg. 18
 b. Distribution of Age According to Disease
c. Is the Disease Due to Family Genes? (Family History Vs Count)
 Family History=Yes Vs Count
 Family History=No Vs Count
2. Correlation heatmap of complete dataset.
3. Machine Learning – Classification (Define and design optimal model by using Naïve
Bayes classification model)
a. Building Feature set
b. Naïve Bayes model
4. Training the model using the pre-trained models which were trained and tested using
the dataset.
5. Testing the model
6. Checking result and Drawing inferences

pg. 19
 CHAPTER 5
DATA EXPLORATION
5.1. Dermatology dataset
The data we handle is Dermatology dataset. This dataset consists of 366 persons data of the
skin diseases categorized into 6 different classes. The dataset consists of 35 columns in total
and here we have 34 columns which are attributes based on which we can determine the
“target” column which represents the type of skin disease occurred. Some of the important
attributes are “age”, “family_history” and “target” (which has numbers ranging from 1 to 6
each representing different type of skin disease).
The entire dataset is being operated to train the model. Later this dataset is split into training
set and testing datasets. Testing set will calibrate the parameters and is used only to get the
performance and efficiency of the system.
5.1.1. Data Manipulation
Manipulation of data is the process of manipulating or changing the information to make it
more structured and understandable to work easily on the data. We use Data manipulation
language (DML) to accomplish this task.
Here are the steps that are to be considered to get commenced with data manipulation:
1. Data manipulation is only possible when you have data to do so. Therefore, you need
a database which is generated from various data sources.
2. Manipulation of data helps in cleaning the information. This work requires the data in
database to be re-organized and re-structured.
3. First of all, we need to import the database to perform the work on the data.
4. We can also insert, remove, and make changes in the databases and its information
with the help of data manipulation step.
5. One of the important steps of our project i.e., data analysis becomes simple after
performing the data manipulation step.

pg. 20
The information or data is in different formats (like .csv, .txt, .xlsx) so it is converted into a
standard format that is all the data is converted in .csv format.
5.1.2. Data Preparation
Data preparation is the one of the important steps to be done before we analyse the data. It is
the process of cleaning the raw data available in the dataset before processing and analysis. It
also involves reformatting, correcting mistakes and the consolidating the data sets to polish
data.
5.1.2.1. Missing Values
We should check whether there are missing values in any of the columns and should handle
them properly.
Generally missing values are either filled with random values leading to less accuracy or they
are ignored which is not good if half of the data contains missing values. The efficient way to
fill these missing values is by finding the correlation between the other attributes.

Figure 5.1: Checking for number of missing values in each column

5.1.2.2. Duplicate Values
We should check whether there are duplicate or repeated values in any of the rows and should
handle them properly.
Generally, it is good to remove the duplicate rows which increases accuracy and saves time.

pg. 21
 Figure 5.2: Checking for number of duplicate rows in dataset

5.1.2.3. Outliers
Outliers are the abnormal data points which don't belong to a particular data population. Their
values lie far away from other values. An outlier simply depicts the mistakes made in
measurements. An outlier is an observation that differ from the well-structured and well-
organized data. We use boxplots, histograms and many other visualisation techniques to find
outliers in the numeric data. In the case of categorical data, it is highly impossible for
outlier detection.

Figure 5.3: Check and remove outliers in age column with histogram

Here we are removing an outlier in age column by checking if there is any person of age<0
which is impossible and is taken as an abnormal case.

Figure 5.4: Check and remove outliers like special chars in age column
Here we are removing an outlier in age column by checking if there is any person of age
containing special character “?” which is impossible for analysis of skin disease and is taken
as an abnormal case.

pg. 22
 CHAPTER 6
DATA ANALYSIS
6.1. Exploratory data analysis (EDA) and its types
Exploratory data analysis (EDA) is a method to analyse the data sets and acquire the
summary of important attributes and their relationships through various forms and techniques
of data visualization based on some statistical data calculated.
These summaries are of 2 types:
1. Numerical: These summaries will be in the form of numbers.
Ex: Average (Mean), Mode, Median, Summation etc.,
Depending on the number of variables, numerical summary is of three types:
 Univariate
 Bivariate
 Multivariate
2. Graphical: Here the summaries are represented by graphs.
Ex: Counter plot, bar graph, histogram, etc.,
6.2. Why do we need Data Analysis?
It is important to analyse the data for the below reasons:
 To identify the distribution of dataset.
 Selecting the appropriate machine learning model and algorithm.
 Best features extraction.
 Evaluation of the model and presentation of the test results.
6.3. EDA Inferences
6.3.1. Count of Different target values (Skin Disease Vs Count)

pg. 23
As we mentioned earlier, there are 6 different types of values in “target” each representing 6
different types of diseases. We plot a counter plot showing the count of number of persons
having each type of skin disease.

Figure 6.1: Counterplot for Skin disease Vs Count

6.3.2. Distribution of Age according to Disease
As we mentioned earlier, there are 6 different types of values in “target” each representing 6
different types of diseases. We plot a Kernel Density Estimate (KDE) plot showing the
distribution of age of persons having each type of skin disease.

Figure 6.2: Distribution of Age according to the Skin disease

6.3.3. Itching factor Vs Skin disease

As we mentioned earlier, there are 6 different types of values in “target” each representing 6
different types of diseases. We plot a Kernel Density Estimate (KDE) plot showing the
distribution of itching factor of persons having each type of skin disease.

pg. 24
 Figure 6.3: Itching factor Vs Skin disease

6.3.4. Family history Vs Count

We plot a counter plot showing the number of persons having skin disease due to family
history and number of people having skin disease not because of family history.

Figure 6.4: Family history Vs Count

6.3.4.1. Family history-YES Vs Count of each disease

We plot a counter plot showing the number of persons having different types of skin diseases
due to family history.

pg. 25
 Figure 6.5: Family history-yes Vs Count

6.3.4.2. Family history-NO Vs Count of each disease

We plot a counter plot showing the number of persons having different types of skin diseases
but are not due to family history.

Figure 6.6: Family history-no Vs Count

6.4. Correlation Heatmap

Correlation heatmaps are the plots that are used to show the visualization of the strength of
the relationship between numerical variables. These plots are also used to estimate which

pg. 26
variables are related to each other and how strong is their relationship. A correlation plot can
find both linear and nonlinear relationships.
Correlation plots contain rows and columns. The columns contain the names of numerical
variables and the rows depicts the relationships between all pairs of variables. Each value in
each cell represents the strength of the relationships.
These relationships are of two types:
1. Positive relationship – If the values are positive
2. Negative relationship – If the values are negative
Below is a diagram showing the correlation heatmap to understand the relationships and their
strength between each variable to all other variables in the dataset by using color-coding
technique. The color-coding technique helps to easily identify the variables and their
relationships just by a glance. In the color scale dark green represents strong relation between
variables and dark red represents weak relations.

pg. 27
 Figure 6.7: Correlation Heatmap

CHAPTER 7
MODELLING
7.1. Model Development
7.1.1. Multi-class classification problem
Multiclass or multinomial classification is a problem which classifies the instances into one
of the three or more classes (classification of instances into one of the two classes is
called binary classification).
While there are many classification algorithms like multinomial logistic regression naturally
permit the use of more than two classes, some are by nature binary algorithms. These
algorithms can, however, be turned into multinomial classifiers by several number of
strategies.
7.1.2. Transformation to binary
To reduce the problem of multiclass classification to multiple binary classification problems
there are some strategies. It can be categorized either into one vs rest or one vs one. These
techniques are developed on the basis of reduction of the multi-class problem into multiple
binary problems. These techniques are also called as problem transformation techniques.
One-vs.-rest
One-vs.-rest (OVR) is a strategy of training a single classifier per class, with the samples of
that class as positive and negative samples. Rather than just a class label, this requires the
base classifiers to produce a real-valued confidence score for making decisions; discrete class

pg. 28
labels alone can cause inconsistencies, whereas we can predict multiple classes for a single
sample.
Here is a pseudocode of the training algorithm for an OVR learner built from a binary
classification learner L:
Inputs:
 L is a learner (training algorithm for binary classifiers)
 X is samples
 y labels where yi ∈ {1, … K} represents the label for the sample Xi
Output:
 fk is a list of classifiers where k ∈ {1, …, K}
Procedure:
 For each value of k in {1, …, K}
 We need to construct a new label vector z where,
If yi = k, then zi = yi else it is 0
 Apply L to X, z so as to obtain the list of classifiers fk
Corresponding classifier reports the highest confidence score for which making decisions
means applying all classifiers to an unseen sample x and predicting the label k.
One-vs.-one
One-vs.-one (OVO) reduction is a strategy of training K(K-1)/2 binary classifiers to reduce
the problem of K-way multiclass classification. From the original training set, each classifier
receives the pair of class samples and also must learn to differentiate between these two
classes. At the time of prediction, we prefer voting. In this voting, these K(K-1)/2 classifiers
are assigned to an unknown and unseen sample and finally the class which got the highest
number of "+1" predictions is predicted by the combined classifier.

7.1.3. Extension from binary

To address the problem of multi-class classification, many algorithms have been evolved on
the basis of decision trees, naïve bayes, k-nearest neighbors, support vector machines, neural
networks and extreme machines for learning. We use Naïve Bayes for the creation of our
model.

pg. 29
7.1.3.1. Naive Bayes
Naive Bayes is one of the successful classifiers which is based on MAP (Maximum a
posterior) estimation.
The Naïve Bayes algorithm contains two words Naïve and Bayes, which are described as
below:
Naïve: It is referred as Naïve, because it assumes in the way for certain features occurrence is
independent of other features occurrences. Such as if the fruit is recognized on the bases of
colour if it red, and taste if it is sweet and shape if it is spherical, then that fruit is recognized
as an apple. Hence every feature contributes individually it’s part to identify whether it is an
apple or not, without relying on other features.
Bayes: As it depends on the law of Bayes’ theorem, it is known as Bayes.
Bayes' Theorem:
Bayes' theorem is also known as Bayes' rule or Bayes' law. It depends on the conditional
probability. Bayes’ theorem is helpful to determine the probability of a hypothesis with
earlier knowledge. The formula for Bayes' theorem is:

Here,
P(A|B): Posterior probability which means it is the probability of hypothesis A on the
observed event B.
P(B|A): Likelihood probability which means it is the probability of the evidence given that
the probability of a hypothesis is true.
P(A): Prior Probability which means it is the probability of hypothesis before observing the
evidence.
P(B): Marginal Probability which means it is the probability of Evidence.
7.2. Model Evaluation
We have created the model and trained it using Naïve Bayes classifier. Now, our task is to
test the testing dataset and predict different metrics like precision score, accuracy, re-call, f1
score and ROC-Curve to determine how well our model is being worked.

pg. 30
 CHAPTER 8
RESULTS AND CONCLUSIONS
8.1. Results
1. We can see that age is the most important factor affecting the skin disease. In our
project we can determine that the persons between age group 10-30 are having high
chance of getting pityriasis_rubra.
2. There are nearly 40 people who suffered from skin disease due to family history and
more than 300 people who suffered from disease but not due to family history.
3. The people having a background of family history are most likely to suffer from the
diseases like pityriasis_rubra, seoboriec_dermatitis, psoriasis.
4. Based on the count of different target values, we can conclude that most of the people
suffered from the disease named “psoriasis”.
5. The precision, re-call and f1 scores obtained for the 6 classes in our project are shown
in the below table:
Table 8.1: Results of Performance metrics of each class label

pg. 31
 Precision Recall F1 Score
Class 0 0.971 0.971 0.971
Class 1 1.0 0.864 0.927
Class 2 1.0 1.0 1.0
Class 3 0.526 0.909 0.667
Class 4 1.0 0.923 0.96
Class 5 1.0 1.0 1.0

8.2. Conclusion
In our project, we have predicted and classified skin diseases using deep learning algorithms.
It is found that we can combine various features and deep learning algorithms to achieve high
accuracy rate and we can also combine various models done previously for the prediction of
many other diseases. The previous models were able to report six skin diseases with
maximum accuracy of 75%. With the usage of deep learning algorithms, we can predict skin
diseases with more accuracy i.e.,99.31%. This shows that deep learning algorithms have a
great potential in the real-world diagnosis of skin disease. This work can be extended in
future by automating the skin disease diagnosis process as it will save time and cost.

REFERENCES
[1] Y. Duan, G. Fu, N. Zhou, X. Sun, N. C. Narendra and B. Hu, "Everything as a Service
(XaaS) on the Cloud: Origins, Current and Future Trends," 2015 IEEE 8th International
Conference on Cloud Computing, 2015, pp. 621-628, doi: 10.1109/CLOUD.2015.88.
https://ieeexplore.ieee.org/document/7214098
[2] Nawal Soliman ALKolifi ALEnezi, A Method Of Skin Disease Detection Using Image
Processing And Machine Learning, Procedia Computer Science, Volume 163, 2019, Pages
85-92, ISSN 1877-0509, https://doi.org/10.1016/j.procs.2019.12.090 (A Method Of Skin
Disease Detection Using Image Processing And Machine Learning - ScienceDirect )
[3] R. Sumithra, Mahamad Suhil, D.S. Guru, Segmentation and Classification of Skin
Lesions for Disease Diagnosis, Procedia Computer Science, Volume 45, 2015, Pages 76-85,
ISSN 1877-0509,
https://doi.org/10.1016/j.procs.2015.03.090
(https://www.sciencedirect.com/science/article/pii/S1877050915003269)

pg. 32
[4] Jaychandra Reddy, V., & Nagalakshmi, T. J. (2019). Skin disease detection using
artificial neural network. Indian Journal of Public Health Research and Development, 10(11),
3829–3832. https://www.ijitee.org/wp-content/uploads/papers/v8i12/L39481081219.pdf
[5] Anurag Kumar Verma, Saurabh Pal, Surjeet Kumar, Comparison of skin disease
prediction by feature selection using ensemble data mining techniques, Informatics in
Medicine Unlocked, Volume 16, 2019, 100202, ISSN 2352-9148,
https://doi.org/10.1016/j.imu.2019.100202
(https://www.sciencedirect.com/science/article/pii/S2352914819300838)
[6] Singhal, E., & Tiwari, S. (2017). Skin Cancer Detection using Arificial Neural
Network. International Journal of Advanced Research in Computer Science, 6(1), 149-157.
doi:https://doi.org/10.26483/ijarcs.v6i1.2402
[7] Hashmani MA, Jameel SM, Rizvi SSH, Shukla S. An Adaptive Federated Machine
Learning-Based Intelligent System for Skin Disease Detection: A Step toward an Intelligent
Dermoscopy Device. Applied Sciences. 2021; 11(5):2145.
https://doi.org/10.3390/app11052145
[8] Shashi Rekha G, Prof. H. Srinivasa Murthy, Dr. Suderson Jena et al. “Digital
Dermatology – skin Disease Detection Model Using Image Processing. Published in
International Journal of Innovative Research in Science, Engineering and Technology. Vol 7,
Issue 7, July 2018. http://www.ijirset.com/upload/2018/july/7_1_DIGITAL.pdf
[9] Ambad, Pravin & Shirsat, A.. (2016). A Image analysis System to Detect Skin Diseases.
IOSR Journal of VLSI and Signal Processing. 06. 17-25. 10.9790/4200-0605011725.
https://www.researchgate.net/publication/
326753284_A_Image_analysis_System_to_Detect_Skin_Diseases
[10] Wei LS, Gan Q, Ji T. Skin Disease Recognition Method Based on Image Color and
Texture Features. Comput Math Methods Med. 2018 Aug 26;2018:8145713. doi:
10.1155/2018/8145713. PMID: 30224935; PMCID: PMC6129338.
https://pubmed.ncbi.nlm.nih.gov/30224935/
[11] R. Yasir, M. S. I. Nibir, and N. Ahmed, “A skin disease detection system for financially
unstable people in developing countries,” Global Science and Technology Journal, vol. 3, no.
1, pp. 77–93, 2015.
https://www.academia.edu/20142809/A_Skin_Disease_Detection_System_for_Financially_U
nstable_People_in_Developing_Countries

pg. 33
[12] Bhadula, S., Sharma, S., Juyal, P., & Kulshrestha, C. (2019). Machine Learning
Algorithms based Skin Disease Detection. International Journal of Innovative Technology
and Exploring Engineering, 9(2), 4044–4049. https://doi.org/10.35940/ijitee.b7686.129219
[13] R. Yasir, M. A. Rahman and N. Ahmed, "Dermatological disease detection using image
processing and artificial neural network," 8th International Conference on Electrical and
Computer Engineering, 2014, pp. 687-690, doi: 10.1109/ICECE.2014.7026918.
https://ieeexplore.ieee.org/abstract/document/7026918
[14] El Abbadi, N. K., Dahir, N. S., AL-Dhalimi, M. A. & Restom, H. (2010). Psoriasis
Detection Using Skin Color and Texture Features. Journal of Computer Science, 6(6), 648-
652. https://doi.org/10.3844/jcssp.2010.648.652
[15] Suneel Kumar and Ajit Singh. Image Processing for Recognition of Skin
Diseases. International Journal of Computer Applications 149(3):37-40, September 2016.
https://www.ijcaonline.org/archives/volume149/number3/25980-2016911373
[16] Połap D, Winnicka A, Serwata K, Kęsik K, Woźniak M. An Intelligent System for
Monitoring Skin Diseases. Sensors (Basel). 2018;18(8):2552. Published 2018 Aug 4.
doi:10.3390/s18082552 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6111999/
[17] Mahamudul Hasan, Surajit Das Barman, Samia Islam, and Ahmed Wasif Reza. 2019.
Skin Cancer Detection Using Convolutional Neural Network. In Proceedings of the 2019 5th
International Conference on Computing and Artificial Intelligence (ICCAI '19). Association
for Computing Machinery, New York, NY, USA, 254–258.
https://doi.org/10.1145/3330482.3330525
[18] M. Shamsul Arifin, M. Golam Kibria, A. Firoze, M. Ashraful Amini and Hong Yan,
"Dermatological disease diagnosis using color-skin images," 2012 International Conference
on Machine Learning and Cybernetics, 2012, pp. 1675-1680, doi:
10.1109/ICMLC.2012.6359626. https://ieeexplore.ieee.org/abstract/document/6359626
[19] V.R. Balaji, S.T. Suganthi, R. Rajadevi, V. Krishna Kumar, B. Saravana Balaji, Sanjeevi
Pandiyan, Skin disease detection and segmentation using dynamic graph cut algorithm and
classification through Naive Bayes classifier, Measurement, Volume 163, 2020, 107922,
ISSN 0263-2241,
https://doi.org/10.1016/j.measurement.2020.107922
(https://www.sciencedirect.com/science/article/pii/S0263224120304607)
[20] https://datahub.io/machine-learning/dermatology

pg. 34
pg. 35
 Appendix: A- Packages, Tools used & Working Process
Python Programming language
Python is a high-level Interpreter based programming language used especially for general-
purpose programming. Python features a dynamic type of system and supports automatic
memory management.
It supports multiple programming paradigms, including object-oriented, functional and
Procedural and also has its a large and comprehensive standard library. Python is of two
versions. They are Python 2 and Python 3.
This project uses the latest version of Python, i.e., Python 3. This python language uses
different types of memory management techniques such as reference counting and a cycle-
detecting garbage collector for memory management. One of its features is late binding
(dynamic name resolution), which binds method and variable names during program
execution.
Python's offers a design that supports some of things that are used for functional
programming in the Lisp tradition. It has vast usage of functions for faster results such as
filter, map, split, list comprehensions, dictionaries, sets and expressions. The standard library
of python language has two modules like itertools and functools that implement functional
tools taken from Standard machine learning.
Libraries
NumPy
Numpy is the basic package for scientific calculations and computations used along with
Python. NumPy was created in 2005 by Travis Oliphant. It is open source so can be used
freely. NumPy stands for Numerical Python. And it is used for working with arrays and
mathematical computations.
Using NumPy in Python gives you much more functional behavior comparable to MATLAB
because they both are interpreted, and they both allows the users to quickly write fast
programs as far as most of the operations work on arrays, matrices instead of scalars. Numpy
is a library consisting of array objects and a collection of those routines for processing those
arrays.
Numpy has also functions that mostly works upon linear algebra, Fourier transform, arrays
and matrices. In general scenario the working of NumPy in the code involves searching, join,
split, reshaping etc. operations using NumPy.

pg. 36
The syntax for importing the NumPy package is → import NumPy as np indicates NumPy is
imported alias np.

Pandas
Pandas is used whenever working with matrix data, time series data and mostly on tabular
data. Pandas is also open-source library which provides high-performance, easy-to-use data
structures and data analysis tools for the Python programming language.
This helps extremely in handling large amounts of data with help of data structures like
Series, Data Frames etc. It has inbuilt methods for manipulating data in different formats like
csv, html etc.,
Simply we can define pandas is used for data analysis and data manipulation and extremely
works with data frames objects in our project, where data frame is dedicated structure for
two-dimensional data, and it consists of rows and columns similar to database tables and
excel spreadsheets.
In our code we firstly import pandas package alias pd and use pd in order to read the csv file
and assign to data frame and in the further steps. We work on the data frames by
manipulating them and we perform data cleaning by using functions on the data frames such
as df.isna().sum(). So, finally the whole code depends on the data frames which are to be
acquired by coordinating with pandas. So, this package plays a key role in our project.
Matplotlib
Matplotlib is a library used for plotting in the Python programming language and it is a
numerical mathematical extension of NumPy. Matplotlib is most commonly used for
visualization and data exploration in a way that statistics can be known clearly using different
visual structures by creating the basic graphs like bar plots, scatter plots, histograms etc.
Matplotlib is a foundation for every visualizing library and the library also offers a great
flexibility with regards to formatting and styling plots. We can choose freely certain
assumptions like ways to display labels, grids, legends etc.
In our code firstly we import the matplotlib.pyplot alias plt, This plt comes into picture in the
exploratory data analysis part to analyze and summarize datasets into visual methods, we use
plt to add some characteristics to figures such as title, legends, labels on x and y axis as said
earlier ,to understand more clearly we can also use different plots.
Seaborn
Seaborn is used for drawing attractive statistical graphics with just a few lines of code. In
other words, we can say seaborn is a data visualization library based on the matplotlib and

pg. 37
closely combined with Pandas data structures in Python. Visualization is the central theme of
Seaborn which helps in exploration and understanding of data.
Plots are used for visualizing the relationship between variables. Those variables can be
numerical or categorical.
Using Seaborn, we can also plot wide varieties of plots like Distribution plots, Pie chart and
bar chart, Scatter plots, Pair plots, Heat maps.
In our code we use seaborn library in EDA where sns is used to create countplot between skin
disease and count of target values and we use facetgrid with respect to sns for looking out
distribution of age based on the diseases and sns with respect to the countplot is also used to
find perspective of data analysis i.e., is the disease due to family genes i.e., family history vs
count. As said earlier, sns also used in determining the heatmap.
Sklearn
Scikit-learn (Sklearn) is the most useful and robust library for machine learning in Python.
Scikit learn is an efficient, and beginner, user friendly tool for predictive data analysis and it
provides a selection of tools for machine learning and statistical modeling including
classification, regression, clustering and dimensionality reduction via a consistent interface in
Python. This library is built upon different libraries such as NumPy, SciPy and Matplotlib.
Scikit learn is used when identifying to which category an object likely belongs, predicting
continuous values and grouping of similar objects into clusters.
When coming to our code, sklearn plays an important with respect to classification algorithm,
the final result and performance. Accuracy of the algorithm can be determined using sklearn.
The different modules imported from sklearn library is train_test_split, GaussianNB, one vs
rest classifier and all the metrics. When going much detail into the modules and packages,
train_test_split means it splits the data into random training and testing subsets. We used
gaussian naive bayes classification algorithm in order to classify the values of the model. On
taking the syntax as example from sklearn. metrics, import * describes to import all the
metrics required for doing some kind of mathematical, or evidential calculations. Similarly
from sklearn.model_selection, import train_test_split described above and there are few
preprocessing steps such as from sklearn.preprocessing import LabelEncoder where
labelencoder encode labels with a value between 0 and n_classes-1 where n is the number of
distinct labels, and other step is from sklearn.preprocessing import label_binarize where label
binarize is used to convert multi-class labels to binary labels (belong or does not belong to the
class) and we use multiclass classification in our which will be explained in detail in the
above document, and when coming to metrics we use confusion matrix in order to calculate

pg. 38
the performance of classification model by using certain measures like precision,recall,f1
sore and threshold value.
Supervised Learning algorithms − Supervised learning is one of the machine learning
approaches through which models are trained using perfectly labelled training data and on the
basis of that models predicts the output. Almost all the popularly known supervised learning
algorithms, Such as Linear Regression, Support Vector Machine (SVM), Decision Tree,
Naïve bayes etc., are the part of scikit-learn.
Unsupervised Learning algorithms − Unsupervised learning is also one of the machine
learning approaches, through which models are not supervised using training data. Instead of
that model itself finds the hidden patterns and insights from the given data.
On the other side it also has all the popular unsupervised learning algorithms from clustering,
factor analysis, PCA (Principal Component Analysis) to unsupervised neural networks.
Clustering − This model can be used to group unlabelled data.
Cross Validation −This process is used to check accuracy on unseen data in supervised
models.
Dimensionality Reduction – Dimensionalities are nothing but attributes of the data. This
step helps in reducing the number of attributes in data which can be used further for tasks like
feature selection, visualization and summarization.
Ensemble methods – Ensemble means to combine. These methods combine various
predictions of multiple supervised models.
Feature extraction – This step is used to define attributes by extracting the features from the
dataset having data of any form.
Feature selection – The extracted features contain lots of features some of which may be not
useful. Feature selection is the process of identifying the important features for the creation of
supervised models.
Open Source − It is an open-source library and also commercially usable under BSD license.
Tools Used Jupyter Notebooks:
Jupyter Project is spin-off project from the I-Python project, which is initially provided an
interface only for the Python language. The name Jupyter is derived from the combination of
Julia, Python, and R.
A Jupyter Notebook is basically a JSON file with a number of annotations. There are mainly
three parts of the Notebook as follows.
 Metadata: A data dictionary of definitions used to set-up and display the notebook; it is
also be said as data about data.

pg. 39
 Notebook format: the Version numbers of the software used to create the notebook. The
version number is used for backward compatibility.
 List of cells: there are three different types of different cells listed beside — markdown
(display), code (to excite), and output.

pg. 40
 Appendix: B
Sample Source Code with Execution

pg. 41
 Paper Publications (if any)
Attach the complete published paper

pg. 42