Moon Analyzer

• We aim to develop a comprehensive tool for visualizing data derived from a machine learning model that classifies X-ray bursts
• The tool will provide both a standalone application and a web-based interface to effectively analyze and interpret the model’s predictions

---

Index

1. Requirement and Deployment
2. Features
3. Preview
4. ML Model
5. ML-model limitations

Requirement and Deployment

Required Modules

Library Name	Version
Python	3.8
PostgresSQL	xx
node.js	20
tailwind CSS	xx

Local Installation

Type the commands inside

1. Check version of Node.

node -v

2. If Node is not available download it for here https://nodejs.org/en/download/ for linux.
3. In the terminal clone the project.

git clone https://github.com/Abhishek-Punhani/Astro-ML-PS

4. Navigate to frontend folder

cd ./Frontend

5. Install node modules in the client.

npm install

6, Navigate to auth and calc folder inside backend folder and install all the dependencies by running this in each folder

pip install -r requirements.txt

6. Start The server separately in each folder

./start_servers.sh

Features

• Interactive Visualizations of graph with different parameters
• Secure login sysem - OTP verification with TLS secured e-mails
• Users are allowed to save projects and can revisit thier analysis without wasting time
• Faster and Lighter with Client-Side Caching and load balancing via Nginx

Preview

Below are the images showing the different graphs and features of the web app



Application of ML Model

Overview

1. The tool processes X-ray light curves and identifies key properties of the detected bursts, such as rise time, decay time, prominence, and peak times. It utilizes clustering methods to classify bursts based on these features.
2. The tool can handle input files in various formats, including ASCII, FITS, and XLS. The classification criteria is based on rise_time, decay_time along with prominenceand the solution is designed to minimize false alarms while maximizing true positive detections.

Features

1. INPUT FORMAT: The tool accepts input files in multiple formats like ASCII, FITS, and XLS.
2. PEAK DETECTION AND CLUSTERING: identify solar flares and fit them to a curve and cluster similar solar flares.
3. OPEN SOURCE: the tool is developed using Python, leveraging libraries such as NumPy, SciPy, scikit-learn, and matplotlib for data processing and analysis.

Limitations of ML Model

1. Sensitivity to Noise:​
   • Smoothing with Gaussian filters (or averaging point smoothing) may not be optimal for all datasets.​
   • Impact: Fine details of smaller bursts might get lost, or noise might still be present in high-intensity bursts.​
2. DBSCAN Clustering Limitations:​
   • DBSCAN may not detect all types of burst patterns, especially if the burst distribution is non-uniform or if noise is high.​
   • Highly sensitive to the eps and min_samples parameters.​
   • Impact: Clustering results may vary significantly based on parameter tuning, and some bursts may be misclassified or marked as outliers.​

​