The Great Space Race to Save the Earth

A Near Earth Object Classification project by Brandon Navarrete

🌎 Goal

Humanity has forgotten how fragile we live. It would be wise to keep an eye to the stary abyss and determine if a near earth object has the potential to be hazardous. With world climate becoming ever changing, we have lost sight on protecting the one resource we all share. EARTH.

• Here I will develop a model that can classify hazardous asteroids given their respective features of diameter, magnitude, and velocity

• This will be encompassed in a report that is easy to read and interpret to any viewers.

🗺️ Data Overview

• This data was pulled from kaggle(2023) which has been pulled from NASA's API

• 90836 rows, each it's own object or asteroids with 10 columns of its features

Initial Questions

• How Many of our Objects Are Inert?

• Will Diameter play a Big Difference in Determining Hazard Status

• Will Relative Velocity play a Big Difference in Determining Hazard Status

• Will Absolute Magnitude play a Big Difference in Determining Hazard Status

Data Dictionary

📂 Data Dictionary

Variable	Value	Meaning
ID	numerical	Unique Identifier for each Asteroid
Name	string	Name given by NASA
est_diameter_min	Float	Minimum Estimated Diameter in Kilometeres.
est_diameter_max	Float	Maximum Estimated Diameter in Kilometeres.
relative_velocity	Float	Velocity Relative to Earth
orbiting_body	string	Earth
sentry_object	False	Included in sentry - an automated collision monitoring system
absolute_magnitude	Float	Describes intrinsic luminosity
Hazardous	Boolean	Feature that shows whether asteroid is harmful or not

Project Plan / Process

1️⃣ Data Acquisition

Gather data from kaggle database

• Import csv in local files

• Read/ Creat data dictionary and extract meaningful columns

acquire.py

• Create acquire.py and user-defined function to import data from csv

2️⃣ Data Preparation

Data Cleaning

• Missing values:

  • No missing values in kaggle dataset

• Outliers

  • Outliers were kept

• Droppeds

  • id,name,orbiting_body, sentry` columns were dropped,no useful information.

Data Splitting

• Create function to split data into train, validate, test

• Call the function, and store the 3 data samples separately in the form of dataframe

3️⃣ Exploratory Analysis

• Ask questions to find what are the key features that are associated with hazard status

• Explore each feature's correlation with status

• Using visualizations to better understand the relationship between features

4️⃣ Statistical Testing & Modeling

• Conduct mann whitney test

• Conclude hypothesis and address the initial questions

5️⃣ Modeling Evaluation

• Find the amount of features that can gerenate the highest performance (Recall)

• Generate XGboost, fit and tranform the train dataset into feature

• Pick the model with highest accuracy and evaluate on test dataset

🏅 Key Findings

• About 10 % of data was classified as hazardous
• All 3 features above shows promise in determing hazard status
• The best performing model was the XGboost and was able to detect 98% of hazardous asteroids

Recommendation

This model has a high percentage of finding the hazardous asteroids at the cost of a low accuracy, due to the false postitives

• This model should be used UNTIL a better model is developed

# Next Steps

• Use the API to gather more relevant features, try to increase hazardous object capture rate.

• Combine with image recogonition, try to automate process to have 24/7 observation / protection

Steps To Clone:

1. Clone this repo
2. Import NASA's csv
3. Run Notebook

   some dependencies may need to be installed such as 'xgboost'