VSX Data Mining

Social Distancing, Light pollution, bad weather... You can still work on meaningful projects from the comfort of your home. All you need is your computer, an internet connection, and patience. This is the first series of AAVSO data mining projects, focusing on unveiling unknown properties of suspected variable stars. Enjoy!

Questions or need help? Email us!

Data Mining Projects

1. Find periods for stars without known values in VSX.

It is common for stars to be labeled as a specific type of variable, but without any  further information. You can help the community by improving this information where possible especially period. The basic steps for designing your project are laid about below:

 - Pick a variability type. VSX has lots of variables, some of which are irregular and won't have periods. Instead, choose something strictly periodic, such as eclipsing binaries.

- Search VSX for all stars of that type. For instance If you wanted to search for all algol type eclipsing binaries you would use EA%.

- In the period range fields enter 0 and 0, you will get a list of all objects without a known period.

- Find existing data on surveys such as ASAS-SN, ASAS-3, NSVS, etc., all available in the VSX external links.

- The star may have elements listed in the ASAS-SN online catalogue of variable stars. If that is the case, you can revise the star entry adding that information and the proper reference.

- If not, analyze survey data for periodicity. This part will require some knowledge of data analysis which you can do using VStar.  For more infomation please go here.

2) Properly classifying stars with Irregular Variability Types

There are many stars with the irregular classification: L, LB or LC (or L:, LB:, LC:)  Most of these turn out not to be semi-regular once a proper analysis has been done.

- Search VSX for irregular variables (e.g. L%)

- Use VSX external links to download survey data on them.

- Data analysis can be done using VStar to determine proper classification. 

3) Reclassifying stars with 'catch-all' variability types.  

Often surveys have classifications for stars that are found to have a specific type of variability, but no further typing could be attributed. Examples of such classes are MISC, PULS, APER, PER, NSIN or SIN.  

- Search VSX for one of these classes (e.g. SIN%)

- Use VSX external links to download survey data on them.

- Data analysis can be done using VStar to help find classification. 

4) Reclassifying stars with VAR variability type. 

VAR stars have even less known about them than the examples listed in the previous type. These stars are known to be variable, but that's it. These will likely be tougher to reclassify as they could fall into just about any variability type. 

- Search VSX using variabilty type VAR 

- Click on the link to a particular star in your list and then check on available data through the VSX external links dropdown menu.

- Data analysis can be done using VStar to help find classification. 

5) Improving new VSX variables.

VSX has a classification of discoveries called novice discoveries. This is when someone has found a star they believe to be variable, but has only used data they collected. Elements such as period can be greatly improved when this is combined with survey data.

-  Go to the VSX search page and select "Novice discoveries..." from the special searches dropdown menu.

- Click on the link to a particular star in your list and then check on available data through the VSX external links dropdown menu.

- Data analysis can be done using VStar to  improve elements.

6) Checking "Not checked" stars.

 Some stars had no good quality data when the survey was entered into VSX. These stars have been labeled as "Not checked". This means that the star identification might be wrong due to blending or the object might be a spurious variable. These stars should be checked to confirm their variable nature.

- Search VSX only checking the "Not checked" box in the search form. There are currently 4344 such records in our database (November 17, 2020), some of them (the KELT ones) are included in Project #7.

- Click on the link to a particular star in your list and then check on available data through the VSX external links dropdown menu.

- Confirm that the object is variable (or not)

- Data analysis can be done using VStar to help find classification and potentially other star elements.

7) Special project

Finding the offending eclipsing binary

We have added a list of 935 false positives (eclipsing binaries originally suspected as being exoplanets) from the KELT survey to VSX.
176 of them are properly identified but there are 759 that were added as "not-checked". This means that their identities need confirmation.
They are stars with close (in terms of the KELT large photometric aperture) bright companions that have been wrongly identified as the variable because a transit-like signal was detected in their photometry due to blending.
Among those 759, we have 148 objects that are perfectly suited for a novice to work with.
These ones have eventually been identified from follow up data but their coordinates have not been published in the KELT paper. Instead they just provided positions and directions pointing to the correct star. That information can be found in the remarks field of each object's page and you may use it to revise them and provide the corrected parameters.

So we have two projects here:

A) Identifying the eclipsing binary following the remark

This is the easiest and most straightforward project.

1) Go to the VSX search page and unfold the dropdown menu in the Special searches section at the top of the page.

2) Select "Finding the offending eclipsing binary..." and click on Go.

3) Go to the star's detail sheet and look at the remark. Identify the object on an image according to its separation and direction from the contaminating bright star. We recommend to use Aladin to easily find the target.

4) Use VSX external links to go to VizieR and find a star that matches the coordinates determined from the image.

5) Confirm the identification by looking at different surveys photometric data (keep in mind each survey's resolution to check if splitting the two objects is possible at the given separation. More info on surveys resolution in the VSX FAQ page). You can use the VSX external links to find survey data. ZTF DR3 data and TESS data can be used too but keep in mind that you won't be able to identify objects with TESS due to the large pixel size.

6) Once the existence of eclipses is confirmed, revise the entry adding the Gaia EDR3 (J2000.0) position of the star. If you can, combine survey data to confirm/improve the elements. Periods are likely twice the value currently given in VSX. Revise the range/amplitude too. The eclipse duration is underestimated in the source catalogue so you can also determine an improved value.

7) Add other names from VizieR catalogues or SIMBAD.

8) Check VizieR to see if the star is included in the ATLAS catalog of variable stars.
If so, you should replace the current KELT primary name by the ATO identifier, add A. N. Heinze et al. (ATLAS) as discoverers and list their paper in the references.

Reference name:  Heinze, A. N.; et al., 2018, A First Catalog of Variable Stars Measured by the Asteroid Terrestrial-impact Last Alert System (ATLAS)
Reference bibcode:  2018AJ....156..241H

9) Mention all you changed in the Revision comment (and add proper references in the References section)

Read more on how to properly revise or submit objects to VSX:

VSX Manual
VSX FAQ page

B) General: confirmation of all not-checked KELT stars (taking the chance to get used to VSX searches...)

1) Go to the VSX search page and click on the More button twice to get extended search options.


2) Write KELT% in the Name field at the top of the form.

3) Scroll down to find the four different variability classes near the bottom. Check the "Not checked" box and uncheck all the others.

When you get the results:

4) Click on an individual star link to get to its VSX detail sheet. If there is a remark indicating distance and direction to the actual EB, jump back to project A.


These projects are merely a few possibilities of data mining projects you can do using VSX. Feel free to come up with your own or expand on those listed here. If you have questions or need assistance you can ask them on the VSX Forum or on the new Data Mining Forum.