We think of our solar system, with 1 star and 8 planets as the norm. But there are other solar systems out there with very different configurations than that with which we are familiar.
For example, it’s possible to have a multi-star system. Our nearest stellar neighbor, the Alpha Centauri system, includes three stars. This diagram illustrates what a 6-star system looks like, with the stars actually existing in pairs:
It’s also possible to have a planet that orbits more than 1 star, we
call those planets Circumbinary:
Image reference: https://en.wikipedia.org/wiki/Circumbinary_planet
We see some of these depictions in Science Fiction, such as in the
planet Tatooine in Star Wars with its “Twin suns”:
NASA keeps a log of every discovered “exoplanet”. An exoplanet is a planet that is part of a solar system that is not our own.
The original Exoplanets dataset had 313 columns and 5438 observations.
In order to focus on the question of “How is our planet and solar system different than other planets and their solar systems?”, I pared down the dataset to only 16 columns with the same number of observations:
| variable | type | na | na_pct | unique | min | mean | max |
|---|---|---|---|---|---|---|---|
| PlanetName | chr | 0 | 0.0 | 5438 | NA | NA | NA |
| HostName | chr | 0 | 0.0 | 4056 | NA | NA | NA |
| PlanetLetter | fct | 0 | 0.0 | 8 | NA | NA | NA |
| NumberOfStars | int | 0 | 0.0 | 4 | 1.00 | 1.10 | 4.0000e+00 |
| NumberOfPlanets | int | 0 | 0.0 | 8 | 1.00 | 1.77 | 8.0000e+00 |
| CircumbinaryFlag | lgl | 0 | 0.0 | 2 | 0.00 | 0.01 | 1.0000e+00 |
| PlanetOrbitalPeriod_Days | dbl | 237 | 4.4 | 5191 | 0.09 | 82218.21 | 4.0200e+08 |
| PlanetRadius_EarthRadius | dbl | 17 | 0.3 | 1642 | 0.31 | 5.67 | 7.7340e+01 |
| PlanetMass_EarthMass | dbl | 24 | 0.4 | 2435 | 0.02 | 484.48 | 2.3900e+05 |
| PlanetEquilibriumTemp_K | int | 1417 | 26.1 | 1506 | 34.00 | 914.66 | 4.0500e+03 |
| StellarLuminosity_Solar | dbl | 223 | 4.1 | 1919 | -6.09 | -0.11 | 3.8000e+00 |
| StellarEffectiveTemp_K | dbl | 207 | 3.8 | 2144 | 415.00 | 5443.26 | 5.7000e+04 |
| StellarRadius_SolarRadius | dbl | 228 | 4.2 | 436 | 0.01 | 1.53 | 1.0946e+02 |
| StellarMass_SolarMass | dbl | 4 | 0.1 | 239 | 0.01 | 0.96 | 1.0940e+01 |
| StellarAge_Gyr | dbl | 1181 | 21.7 | 592 | 0.00 | 4.37 | 1.4900e+01 |
| SystemDistance_Parsecs | dbl | 21 | 0.4 | 3982 | 1.30 | 701.91 | 8.5000e+03 |
So now we have 16 columns to focus on. Let’s first look at the structure of the data, including how many NA’s we should keep in mind.
We have:
I’ll examine the variables at a solar system level first, then look at planets, stars, and then compare all the variables. Finally we’ll run one statistical test to answer a question about different types of solar systems.
An exoplanet is a planet that is in another system. So let’s start by looking into the makeup of those other solar systems.
These distances are measured in parsecs which are used for distances above 3.26 light years. The sun is only 8.3 light minutes away from the earth, so even though Parsecs would not traditionally be used for such a small distance, the distance from the earth to the sun in parsecs is \(\ 4.84x10^{-6}\).
The closest solar system is 1.3 parsecs away which means it is about 4.2 light years away. And as a reminder, a light year is roughly 5.88 trillion miles!
This chart combines a violin plot that displays the density of the observations with a boxplot.
What are the number of stars and planets in other solar systems? Unlike our own system of only 1 star, some solar systems have two, three, even four stars! And also unlike our own 8-planet system, most other systems have only 1 planet (that we’ve discovered!).
Circumbinary planets orbit two stars, as shown in the diagram in the introduction. How common is this?
The majority of planets are not circumbinary, with only 0.8% of planets in this data set being circumbinary. So now we know that earth is in the majority, with the other 90.2% of planets in this data set that are not circumbinary.
Now we can look at the Planet variables and see how they differ on planets that are and aren’t circumbinary.
For example, here we can see that circumbinary planets tend to have a higher mass than planets that are not circumbinary.
We know that the orbital period of the earth around its sun is 365 days. How does that compare with other suns in our solar system?
We can also look at the relationship between a planet’s orbital period (in days) and it’s Equilibrium Temperature. This looks negatively correlated, which make sense intuitively: As a planet gets further from its sun, it gets colder.
There is a general positive correlation between a planet’s mass and it’s radius.
What are the characteristics of the stars in this dataset?
Some interesting things emerge, such as a strong positive correlation between Mass and Luminosity.
Note: Luminosity is the “Amount of energy emitted by a star per unit time, measured in units of solar luminosities”
Let’s take a look at our sun relative to some of those dimensions - in this case, I’ll pick the strongest positive correlation of Stellar Mass vs. Stellar Luminosity. I also added a linear regression model line marked in red.
A large correlation matrix of all the measured numerical variables for Planets and Stars.
First I started by creating groups that separated solar systems into 2 types: Single star systems and Multi-star systems.
Solar system groups
## Multi-Star Single-Star
## 489 4949
Then I created two equal sized random samples with 400 observations each.
From this I now had a categorical variable to measure against a variety of numeric continuous variables, including:
Let’s focus on Planet Mass.
So from this, I have a hunch that planets in Multi-star Solar Systems have a different mean mass than planets in a Single-star Solar System.
Our confidence interval is 95%, and our signficance level (or alpha) is 0.05.
H0: The true difference between these group means is zero.
H1: The true difference between these group means is not zero.
We will treat the types of solar systems as two different groups/samples. So we’ll use an Independent Samples T-test to test if this hunch is true and statistically significant.
##
## Welch Two Sample t-test
##
## data: sample_multi_star_systems$PlanetMass_EarthMass and sample_single_star_systems$PlanetMass_EarthMass
## t = 4.0449, df = 780.55, p-value = 5.754e-05
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## 162.9601 470.2663
## sample estimates:
## mean of x mean of y
## 664.9497 348.3365
Our p-value of \(5.754e^{-5}\) is smaller than our alpha of 0.05, therefore we can say with 95% confidence that planets in Multi-star systems have a different mean mass than planets in Single-star system.
Some fun facts you can take with you today:
There is so much more to explore! It was fascinating to learn so much about exoplanets.
Other things to dig into: