This repository contains university project. The goal of this project is to create a program which helps people to automatically classificate variable stars or differ other objects to variable stars using machine learning algorithms.
- In progress
- Python
- pandas
- psycopg2 (Postgres SQL)
- csv
- astropy
- astroquery
- Subject area
- Basics of astronomy
- Variable stars classification
- Coordinate systems and units
- Etc
We use database which is the result of crossing APASS DR9 and GALEX AIS GR6+7 by X-Macth.
| Field | Description | Example |
|---|---|---|
| FUVmag | [10.6/23.8] GALEX FUV calibrated magnitude in AB system | 21.8688 |
| NUVmag | [-10.5/4.4] NUV Kron-like elliptical aperture magnitude | 18.1355 |
| nobs | [2/387] Number of observed nights | 8 |
| mobs | [2/3476] Number of images for this field, usually nobs*5 | 40 |
| Vmag | [5.5/27.4] Johnson V-band magnitude. Optical V band between 500 and 600 nm | 13.551 |
| e_Vmag | [0/7] Vmag uncertainty | 0.066 |
| Bmag | [5.4/27.3] Johnson B-band magnitude. Optical B band between 400 and 500 nm | 14.209 |
| e_Bmag | [0/10] Bmag uncertainty | 0.075 |
| gpmag | g'mag [5.9/24.2] g'-band AB magnitude, Sloan filter. Optical B band between 400 and 500 nm | 13.851 |
| rpmag | r'mag [5.1/23.9] r'-band AB magnitude, Sloan filter. Optical R band between 600 and 750 nm | 13.376 |
| ipmag | r'mag [4.2/29.1] i'-band AB magnitude, Sloan filter. Optical I band between 750 and 1000 nm | 13.208 |
| otype | Object type from Simbad | EB* |
| starType | Variability Type, as in GCVS catalog Variability type (see details of VSX type list) | EW |
| min | Magnitude at minimum, or amplitude | 0.10999999940395355 |
| max | Magnitude at maximum, or amplitude | 13.399999618530273 |
| Period | Period of the variable in days | 0.290016 |
Train dataset: (6683, 8) (6683,)
Test dataset: (1671, 8) (1671,)
Nearest Neighbors
precision recall f1-score support
0 0.95 0.97 0.96 818
1 0.97 0.95 0.96 853
accuracy 0.96 1671
macro avg 0.96 0.96 0.96 1671
weighted avg 0.96 0.96 0.96 1671
Linear SVM
precision recall f1-score support
0 0.94 0.97 0.96 811
1 0.97 0.94 0.96 860
accuracy 0.96 1671
macro avg 0.96 0.96 0.96 1671
weighted avg 0.96 0.96 0.96 1671
RBF SVM
precision recall f1-score support
0 0.95 0.97 0.96 816
1 0.97 0.95 0.96 855
accuracy 0.96 1671
macro avg 0.96 0.96 0.96 1671
weighted avg 0.96 0.96 0.96 1671
Decision Tree
precision recall f1-score support
0 0.96 0.97 0.96 828
1 0.97 0.96 0.96 843
accuracy 0.96 1671
macro avg 0.96 0.96 0.96 1671
weighted avg 0.96 0.96 0.96 1671
Random Forest
precision recall f1-score support
0 0.95 0.97 0.96 817
1 0.97 0.95 0.96 854
accuracy 0.96 1671
macro avg 0.96 0.96 0.96 1671
weighted avg 0.96 0.96 0.96 1671
Neural Net
precision recall f1-score support
0 0.94 0.98 0.96 801
1 0.98 0.94 0.96 870
accuracy 0.96 1671
macro avg 0.96 0.96 0.96 1671
weighted avg 0.96 0.96 0.96 1671
AdaBoost
precision recall f1-score support
0 0.95 0.97 0.96 821
1 0.97 0.95 0.96 850
accuracy 0.96 1671
macro avg 0.96 0.96 0.96 1671
weighted avg 0.96 0.96 0.96 1671
Naive Bayes
precision recall f1-score support
0 0.94 0.97 0.95 811
1 0.97 0.94 0.95 860
accuracy 0.95 1671
macro avg 0.95 0.95 0.95 1671
weighted avg 0.95 0.95 0.95 1671
QDA
precision recall f1-score support
0 0.94 0.97 0.95 806
1 0.97 0.94 0.96 865
accuracy 0.96 1671
macro avg 0.96 0.96 0.96 1671
weighted avg 0.96 0.96 0.96 1671
Best accuracy: Decision Tree 0.9622980251346499