Scribd
Upload
100 views15 pages

Final Year Project: Asteroids Classification Using Machine Learning

The document describes a final year project on classifying asteroids as hazardous or non-hazardous using machine learning. It was submitted by 5 students and supervised by a professor. The project involved collecting asteroid data, preprocessing it, and using KNN, decision tree, and random forest classifiers to perform the classification. The random forest model achieved the highest accuracy of 99.57% while KNN had the lowest at 83.15%. The conclusion is that random forest and decision tree produced the best results for this asteroid classification task.

Uploaded by

kavyaa chakraborty
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
100 views15 pages

Final Year Project: Asteroids Classification Using Machine Learning

The document describes a final year project on classifying asteroids as hazardous or non-hazardous using machine learning. It was submitted by 5 students and supervised by a professor. The project involved collecting asteroid data, preprocessing it, and using KNN, decision tree, and random forest classifiers to perform the classification. The random forest model achieved the highest accuracy of 99.57% while KNN had the lowest at 83.15%. The conclusion is that random forest and decision tree produced the best results for this asteroid classification task.

Uploaded by

kavyaa chakraborty
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 15

Final Year Project

Asteroids classification using machine learning


SUBMITTED BY UNDER SUPERVISION
DEBRAJ DHAR (CSE 4A, ROLL - 83)
OF
TIRTHA MAITRA (CSE 4A, ROLL - 53)
PROF. RATAN KR.
CHAYAN KUMAR KAYAL (CSE 4A, ROLL - 54)

RAJARSHI BOSE (CSE 4A, ROLL - 61) BASAK


SOUGATO BAGCHI (CSE 4A, ROLL - 64)
 Abstract
 Introduction
 Literature survey
 Inspiration

Contents  Experimental methodology


 Results
 Conclusion
 The project is about the classification of potential hazardous or
non-hazardous asteroids.

 The dataset is about Asteroids


Abstract
 With the help of different Machine Learning models and
algorithms, we have conducted a thorough and comparative
analysis of the experiment.
Asteroids:
 Asteroids, sometimes called minor planets, are rocky remnants
left over from the early formation of our solar system about 4.6
billion years ago.
Introduction
 Most asteroids are irregularly shaped, though a few are nearly
spherical, and they are often pitted or cratered.
 Models used:

Introduction  K-Nearest Neighbors

cont.  Decision tree

 Random Forest
 The classification of asteroids

J.K.Davies, N Eaton
Department of Astronomy, University of Leicester, U.K.

Literature  Asteroid families classification: Exploiting very


Survey large datasets

Andrea Milani, Alberto Cellino, Zoran Knežević, Bojan Novaković, Federica Spoto,
Paolo Paolicchi
What is the reason to conduct an experimental analysis on
Inspiration classification of asteroids?
 Data Preprocessing: To ensure no abnormal or inappropriate or
unrelated values in the dataset, some pre-requisite steps were
Experimental performed on the dataset.

Methodology  After the completion of data preprocessing, the dataset was split
into two sets to continue with our experiment.
 Moving on to the next step, we used 3 different machine learning
classification models (K-nearest neighbor classifier, Decision Tree
and Random forest classifier)

Experimental
Methodology  At the end of the experiment, all the models which were used to
cont. evaluate the performance, were put up against each other for a
comparative analysis.
• Figure depicts the plot of the number of hazardous asteroids
compared to non-hazardous asteroids.

Results

Number of Hazardous Asteroids vs Non-Hazardous Asteroids


• The confusion matrix for each of the models are the following figures:

KNN Classifier

Predicted: No Predicted: Yes


Results Actual:No 1118 57
cont. Actual:Yes 180 52

Decision Tree Classifier Random Forest Classifier

Predicted: Predicted: Predicted:


No Yes No Predicted: Yes

Actual:No 1170 5 1171 4

Actual:Yes 3 229 2 230


 Comparative Analysis

Results Random
cont. K-NN Decision Tree Forest

Accuracy (%) 83.15 99.43 99.57

Precision (%) 47.7 97.86 98.29

F-Measure 0.3049 0.9828 0.9871

Recall 0.2241 0.987 0.9913


 From the experimental results, the primary objective of our
experiment has been fulfilled.
 Moreover, we found out that at maximum 2 out of 6 asteroids on
average can be hazardous.
 With the flow of time, the number of asteroids seen on average is
increasing in a steady but slow curve, but a good prediction model
Conclusion can be used to predict that accurately.
 From the comparative analysis we concluded that Random Forest
classifier and Decision Tree Classifier produces the most optimal
result a. On the other hand, K-NN classifier produces the least
desirable outcome.

You might also like