ABSTRACT
Auto-ML sets out as connection joining the different levels of competence when
building Machine learning pipelines or systems and achieve the data science
processes more quickly. We present the common Automl tool which works on
cleaned datasets using normal Ready-made algorithms provided by sklearn to run
against regression and classification datasets and we also use the open source
Automl libraries like auto-sklearn,hyperpot,Tpot etc and found that TPOT is best
suitable for the regression datasets and auto-sklearn is best suitable for the
classification datasets .The auto sklearn library is basically used for running the auto-
ml . By using the scikit-learn machine learning library we can train the high level of
machine learning algorithms and can find best accuracy easily for the given dataset.
v
TABLE OF CONTENTS
ABSTRACT v
LIST OF FIGURES vii
CHAPTER TITLE PAGE
NO. NO.
INTRODUCTION
1 1
1.1 OBJECTIVE OF THE PROJECT
1.1.1 NECESSITY 1
1
1.1.2 SOFTWARE DEVELOPMENT METHOD
1.1.3 LAYOUT OF THE DOCUMENT 2
2
1.2 OVERVIEW OF THE PROJECT
2
2 LITERATURE SURVEY 3
2.1 LITERATURE SURVEY
3
3 METHODOLOGY
5
3.1 PROJECT PROPOSAL
5
3.1.1 MISSION 5
3.1.2 GOAL
3.2 SCOPE OF THE PROJECT 5
5
3.3 OVERVIEW OF THE PROJECT
5
3.4 FLOWCHART
6
4 SYSTEM DESIGN AND IMPLEMENTATION
7
4.1 SYSTEM STUDY
7
4.1.1 SYSTEM REQUIREMENT SPECIFICATIONS
7
4.2 SYSTEM SPECIFICATIONS
4.2.1 MACHINE LEARNING OVERVIEW 7
7
4.2.2 FLASK OVERVIEW 8
vi
8
4.3 PYTHON LIBRARIES NEEDED
4.3.1 NUMPY LIBRARY 8
8
4.3.2 PANDAS LIBRARY
4.3.3 MATPLOTLIB LIBRARY 8
9
4.3.4 SEABORN LIBRARY
4.3.5 SCIKIT LEARN LIBRARY 9
4.4 DESCRIPTION ABOUT EACH LIBRARY 9
4.4.1 HOW NUMPY CAN BE USED 9
4.4.2 HOW PANDAS CAN BE USED 10
11
4.5 MODULES
4.5.1 DATA PRE-PROCESSING 12
4.5.2 DATA VALIDATION /CLEANING /PREPARING 13
PROCESS
4.5.3 EXPLORATION DATA ANALYSIS OF VISUALIZATION 14
4.5.4 COMPARING ALGORITHM WITH PREDICTION IN THE FORM 14
OF BEST ACCURACY RESULT
4.5.5 ALGORITHM AND TECHNIQUES 18
4.6 DEPLOYMENT USING DJANGO 30
4.7 DETAIL EXPLANATION OF AUTO ML
32
vii
5 RESULTS AND DISCUSSION, PERFORMANCE ANALYSIS 33
6 SUMMARY AND CONCLUSION 34
6.1 SUMMARY 34
6.2 CONCLUSION 34
REFERENCES 35
APPENDIX
A. SOURCE CODE 37
B. SCREENSHOTS 42
C. PUBLICATION 46
viii
LIST OF FIGURES
FIGURE NO. FIGURE NAME PAGE NO.
3.1 System Architecture 6
3.2 Flow chart 6
4.2 Logistic Regression 20
4.3 Linear Regression 21
4.4 Random Forest 22
4.5 Decission Trees 24
4.6 Naïve Bayes 25
4.7 K Nearest Neighbor 26
4.8 Support Vector Classifier 27
4.9 Support Vector Regressor 28
4.10 Gradient Boosting 28
4.11 XG Boosting 29
4.12 Adaptive Boosting 30
0
CHAPTER-1
INTRODUCTION
1.1 OBJECTIVE OF THE PROJECT:
To build an Auto-ML tool for any cleaned dataset so that it applies
machine learning algorithm and predicts the accuracy scores for all applied algorithms
and also give top three result.
The recent substantial progress in machine learning has led to a growing demand for
hands-free ML systems that can support developers and ML novices in efficiently
creating the new ML applications. Since different datasets require different ML pipelines,
this demand has given rise to the area of automated machine learning.
1.1.1 Necessity: This website helps in overcoming the time management. This
Application is very easy to use. It can work accurately and very smoothly in a different
scenario. It reduces the effort workload and increases efficiency in work. In aspects of
time value, it is worthy.
In this website the user can easily use our auto-ml tool for choosing the best algorithm
for the given supervised machine learning data. Our auto-ml tool provides the accuracy
scores of all algorithms for classification/Regression data. Then it displays the best three
accuracy models for the uploaded dataset. So, by using our auto-ml tool the user can
easily find the best suitable model for the data
Hence even the non-coders also easily done machine learning by using our tool.
1.1.2 Software development method:
In many software applications program different methods and cases are
followed such as, Waterfall model, Iterative model, Spiral model, V-model and Big Bang
model. I used waterfall model in this application. I tried to use test case and case
software approaches.
1
1.1.3 Layout of the document:
This documentation starts with formal introduction. After introduction analysis
and design of the project are described. In analysis and design of the project have many
parts such as project proposal, mission, goal, target audience, environment. After that
design and table diagram will be found. Use cases and test cases are in chapter 2 and
chapter 3 respectively. Finally this documentation finished with result and Conclusion
part.
1.2 Overview of the Designed Project:
At first we take the dataset from out resource then we have to perform data-
preprocessing, visualization methods for cleaning and visualizing the dataset
respectively and we upload the cleaned dataset and can run all algorithms easily by
clicking button for getting accuracy scores and then it will give best three accuracy
scores algorithms and flask is used for user interface.
2
CHAPTER-2
LITERATURE SURVEY
2.1 LITERATURE SURVEY:
General
A literature review is a body of text that aims to review the critical points of current
knowledge on and/or methodological approaches to a particular topic. It is secondary
sources and discuss published information in a particular subject area and sometimes
information in a particular subject area within a certain time period. Its ultimate goal is to
bring the reader up to date with current literature on a topic and forms the basis for
another goal, such as future research that may be needed in the area and precedes a
research proposal and may be just a simple summary of sources. Usually, it has an
organizational pattern and combines both summary and synthesis.
A summary is a recap of important information about the source, but a synthesis is a re-
organization, reshuffling of information. It might give a new interpretation of old material
or combine new with old interpretations or it might trace the intellectual progression of
the field, including major debates. Depending on the situation, the literature review may
evaluate the sources and advise the reader on the most pertinent or relevant of them.
Review of Literature Survey
Title : Benchmarking Automatic Machine Learning Frameworks
Author: Adithya Balaji, Alexander Allen
Year : 2018
We test auto-sklearn, TPOT, auto_ml, and H2O's AutoML solution against a
compiled set of regression and classification datasets sourced from OpenML and found
that TPOT is best suitable for the regression type datasets and auto-sklearn is best
suitable for the classification type datasets
Title : Efficient and Robust Automated Machine Learning
Author: Matthias Feurer, Aaron Klein, Katharina Eggensperger
Year : 2015
3
The demand for the machine learning has been increased due to the success of
machine learning in various range of applications .For good effective, such systems
need to automatically chooses the algorithm and data pre-processing steps for a new
dataset and should set their respective hyper-parameters.
Title : Hyper-parameter Optimization of Machine Learning Algorithms
Author: Li Yang, Abdallah Shami
Year : 2020
Currently most of us using machine learning algorithms in different applications and
in different areas. For fitting into different problems the machine learning algorithms
have to do hyper-parameters must be tuned. There is an direct impact on models
performances by selecting the best hyperparameter configuration for ML models. It even
also requires more knowledge on ML algorithms and hyper-parameter optimization.
Title : Hyperopt-Sklearn: Automatic Hyperparameter Configuration for Scikit-learn
Author: Brent Komer, James Bergstra, Chris Eliasmith
Year : 2014
Hyperopt-sklearn is the project which gives the auto algorithm configuration for
the sklearn ML library. Following Auto-Weka, To represent a single large hyper-
parameter optimization problem we can take the choice of large pre-processing modules
and the choice of classifier together.
4
CHAPTER-3
METHODOLOGY
3.1 Project Proposal:
The project proposal is the term of documents. A project can describe the
project proposal. It is the set of all plans of a project. Like, how the software works, what
are the steps to complete the entire projects, and what are the software requirements
and analysis for this project. In my project, I am doing all the steps and also risk and
reward and other project dependencies in the project proposal.
3.1.1 Mission:
An online Web based machine learning application is very popular and well
known to everyone. We can train high level custom machine learning models with
minimal effort and machine learning expertise. Hence even non coder also can do
machine learning easily by using our Auto-ML tool. This simple Auto-ML tool gives fast
and accurate results for choosing the best model for the given dataset.
3.1.2 Goal:
The goal is to build an Auto-ML tool for choosing the best accuracy model
for any given cleaned dataset.
3.2 Scope of the Project:
The scope of this paper is to implement and investigate how different
supervised binary classification methods impact default prediction. The model evaluation
techniques used in this project are limited to precision, sensitivity, F1-score.
3.3 Overview of the Project:
The overview of the project is to build an Auto-ML tool for choosing the
best accuracy model for any given cleaned dataset. So we can do any classification and
regression type projects can predict the data and displays the best three accuracy
model.
5
3.4 Flow Chart:
Fig:3.1: Machine Learning workflow diagram
The above flow diagram represents the flow of process from gathering data to
predicting result.
Fig:3.2: System Architecture
Clean the raw data then apply machine learning algorithms and finally predict the
results.
