You are on page 1of 13

Soil Quality

Analysis & crop


suggestions
using
Machine Learning
The Team -

Shraddha Shreedhar Vaibhav Upadhyay


4NM17CS168 4NM17CS205
Dept of CSE Dept of CSE
NMAMIT, Nitte NMAMIT, Nitte

Under the guidance of -


Mr. Sandeep Kumar Hegde
Assistant Professor II
Dept of CSE
NMAMIT, Nitte
TABLE OF CONTENTS

Abstract 01 03 Problem Definition

Introduction 02 04 Literature Survey

Proposed
Architecture 05 06 Conclusion
ABSTRACT
In this project, we propose a novel
method that aims at finding significant
features by applying machine learning
techniques resulting in improving the
accuracy in the prediction of
cardiovascular disease. The prediction
model is introduced with different
combinations of features and several
known classification techniques. This
dataset is a real world data including
important features of patients. We will
build the predictable model by using
various algorithms.
INTRODUCTION

● Heart disease is a cardiovascular disease (CVD) that remains the number one cause
of death globally and contributes to approximately 30% of all global deaths. If
unmitigated, the total number of deaths globally is projected to increase to around
22 million in 2030.

● Medical diagnostic opinion is a prime factor of the clinical subject, considering that
the diagnostic method likely to have comparatively significant ambiguities, which
leaves a scope of improvement with the boom of technological advancement,
mainly in the field of statistical learning.

● The main topic is heatlh grade level prediction using machine learning techniques.
Machine learning is widely used nowadays in various applications. Our topic is about
prediction of heart disease by processing patient’s dataset and a data of patient to
whom we need to predict the chance of occurrence of a heart disease.
LITERATURE SURVEY
PROBLEM DEFINITION

During our literature survey, we identified few of the problems in past


approaches. The algorithms performed according to number of epochs they
were trained upon and the way the raw data was treated. But, there is a thumb
rule of machine learning that any model starts overfitting if we keep running
them for more than required amount of time. This leads us to our first
problem, that is failure of certain learning approaches to generalize. Moving
to our second problem, which in inherent to our dataset that is, really less
number of data points, because of which it is really not sure how the
proposed models will perform in real world scenarios and hence ruling out all
exhaustive algorithms of deep learning from the consideration. So, we
propose a hyperparameter tuned version of the well performing algorithms.
PROPOSED ARCHITECTURE

UCI Heart Disease Dataset

Cleansing & preparation of data

Analysing data & building models

Merging models & accuracy calculation


PROPOSED ARCHITECTURE

The algorithms we will be using are:

● Logistic Regression
● K-Nearest Neighbours Classifier
● Support Vector machine
● Decision Tree Classifier
● Random Forest Classifier
● XGBoost Classifier
Workflow of our
project
CONCLUSION
In future the work can be enhanced by developing a notebook, a web
application based on the these algorithms mentioned and a larger dataset
as compared to the one used in this analysis to make a more industrial
robust ML health grade model, which will help to provide better results and
help health professionals in predicting the heart disease effectively and
efficiently.
THANK YOU

You might also like