Stellar Object Classification: Unveiling the Cosmos

Dive into the fascinating world of astrophysics with our SlideShare presentation on stellar object classification! Explore the criteria and techniques used to categorize celestial bodies, from stars and planets to galaxies and nebulae. We’ll discuss the Hertzsprung-Russell diagram, spectral classification, and the significance of each type of stellar object in our understanding of the universe. Packed with stunning visuals and insightful data, this presentation is ideal for students, educators, and astronomy enthusiasts eager to deepen their knowledge of how we classify the myriad wonders of the cosmos. Join us on this enlightening journey through space! for more details visit: https://bostoninstituteofanalytics.org/data-science-and-artificial-intelligence/

1. CONFIDENTIAL : The information in this document belongs to Boston Institute of Analytics LLC. Any unauthorized sharing of this material is prohibited and subject to legal action under breach of IP and confidentiality clauses. Capstone Project On Stellar Object classification

2. Agenda CONFIDENTIAL : The information in this document belongs to Boston Institute of Analytics LLC. Any unauthorized sharing of this material is prohibited and subject to legal action under breach of IP and confidentiality clauses. Objective of the Project Data Gathering/Data Understanding Exploratory Data Analysis Feuture Selection Categorical Feauture Analysis Numerical Feauture Analysis Data Standardization Train Test Split Modelling

3. Objective: CONFIDENTIAL : The information in this document belongs to Boston Institute of Analytics LLC. Any unauthorized sharing of this material is prohibited and subject to legal action under breach of IP and confidentiality clauses. Develop comprehensive machine learning models to accurately classify celestial objects as stars, galaxies, or quasars. Project Benefits: Automated Classification: The models will help efficiently categorize large numbers of celestial objects from survey data. Research Insights: Understanding the key features that distinguish different celestial objects can guide future astronomical research. Data Processing Optimization: Identifying the most effective classification algorithms can improve data processing pipelines for large- scale astronomical surveys.

4. Data Gathering and Understanding: Shape : (100000, 18) 17 Columns are Numerical 1 Columns is Categorical No Null Data Present Lable- encoding for Caegorical Column CONFIDENTIAL : The information in this document belongs to Boston Institute of Analytics LLC. Any unauthorized sharing of this material is prohibited and subject to legal action under breach of IP and confidentiality clauses.

5. Standardization & Principal Component Analysis! Standardization: To ensure that all features contribute equally to the model, we standardized the dataset by removing the mean and scaling features to unit variance. This step helps in normalizing data that might otherwise have varying scales, allowing algorithms like PCA and machine learning models to perform optimally. PCA was employed as dimensionality reduction technique to simplify the dataset which retaining most of it variance . By projecting the data onto a set of principal component we reduce the number of features, helping to minimise computational complexity without sacrificing the dataset essential information.

6. Distribution of Variables

7. Feauture Variable with the Target Variable

8. Correlation Coefficience we did feature selection by looking the correlation coefficients and we remove 9 of them whose correlation coefficients are between -0.067 and 0.056 CONFIDENTIAL : The information in this document belongs to Boston Institute of Analytics LLC. Any unauthorized sharing of this material is prohibited and subject to legal action under breach of IP and confidentiality clauses.

9. Train and Test Split CONFIDENTIAL : The information in this document belongs to Boston Institute of Analytics LLC. Any unauthorized sharing of this material is prohibited and subject to legal action under breach of IP and confidentiality clauses. We have taken 33% Data to be test and 67% to Train Data. Setting a random state ensures consistent results and using stratify=y maintains a proportional distribution of the target variable in both sets.

10. Logistics Regression CONFIDENTIAL : The information in this document belongs to Boston Institute of Analytics LLC. Any unauthorized sharing of this material is prohibited and subject to legal action under breach of IP and confidentiality clauses.

11. Random Forest CONFIDENTIAL : The information in this document belongs to Boston Institute of Analytics LLC. Any unauthorized sharing of this material is prohibited and subject to legal action under breach of IP and confidentiality clauses.

12. KNN ( K- Nearest Neighbour ) CONFIDENTIAL : The information in this document belongs to Boston Institute of Analytics LLC. Any unauthorized sharing of this material is prohibited and subject to legal action under breach of IP and confidentiality clauses.

13. Support Vector Mechanism Accuracy : 0.972 CONFIDENTIAL : The information in this document belongs to Boston Institute of Analytics LLC. Any unauthorized sharing of this material is prohibited and subject to legal action under breach of IP and confidentiality clauses.

14. Ensemble Modelling CONFIDENTIAL : The information in this document belongs to Boston Institute of Analytics LLC. Any unauthorized sharing of this material is prohibited and subject to legal action under breach of IP and confidentiality clauses.

15. Conclusion CONFIDENTIAL : The information in this document belongs to Boston Institute of Analytics LLC. Any unauthorized sharing of this material is prohibited and subject to legal action under breach of IP and confidentiality clauses. After performance of Model with Using above mentioned techniques, SVM is having more accuracy Using large-scale data and machine learning has greatly improved predicting stars and celestial objects. These accurate predictions help us learn more about space and support exploration. As data science and astronomy advance, we'll get even better at understanding and predicting space events.

16. Thank You!