Welcome to StatKit, a collection of statistical analysis tools for Swift developers.
StatKit adds relevant functionality for statistical analysis for the types you use every day. With StatKit, you will be able to calculate a variety of useful statistics such as:

Central Tendency
Calculate modes, means, and medians of your data sets. 
Variability
Compute variances and standard deviations. 
Correlation
Find linear tendencies and covariance of measurements. 
Distributions
Make computations using several common distribution types.
A simple example would be to calculate the modes of an integer array, which can be done easily with the following piece of code:
print(mode(of: [1, 2, 3, 3, 2, 4], variable: \.self))
// Prints [3, 2]
In this case, mode(of:variable:)
takes a KeyPath argument which specifies the variable inside the array that you are interested in. In the example above, we specify the \.self
keypath, which points to the array element itself (in this case, the integers).
The pattern of specifying one or more variables to investigate is common throughout the StatKit library. It allows you to calculate similar statistics for a variety of different types using the same syntax. For example, both of the below examples produce valid results, even though the types under investigation are completely disparate:
Calculating the mode of all characters in a String
:
print(mode(of: "StatKit", variable: \.self))
// Prints ["t"]
Calculating the mode of CGPoint
yvalues in an array:
import CoreGraphics
let points = [CGPoint(x: 0, y: 1),
CGPoint(x: 1, y: 3),
CGPoint(x: 3, y: 1)]
print(mode(of: points, variable: \.y))
// Prints [1.0]
As the examples in the previous section showed, calculating statistics is easy when using collections of types that are readily available. However, most of us work with custom data structures in our projects. Luckily, StatKit provides support for arbitrary custom types thanks to the extensive use of generics.
Let us look at a custom data structure that keeps track of collected data points for a specific brand of cars, and how we can use StatKit to wasily calculate the mean and standard deviation of their fuel consumption:
struct FuelConsumption {
let modelYear: String
let litersPer10Km: Double
}
let measurements: [FuelConsumption] = [...]
mean(of: measurements, variable: \.litersPer10Km, strategy: .arithmetic)
standardDeviation(of: measurements, variable: \.litersPer10Km, from: .sample)
As you can see, using KeyPath's makes the StatKit API easy to use and reusable across completely arbitrary custom structures.
StatKit provides multiple discrete and continuous distribution types for you to work with. These allow you to compute probabilities, calculate common moments such as the skewness and kurtosis, and sample random numbers from a specific data distribution.
let normal = NormalDistribution(mean: 0, variance: 1)
print(normal.cdf(x: 0))
// Prints 0.5
let normalRandomVariables = normal.sample(10)
// Generates 10 samples from the normal distribution
StatKit is documented using SwiftDocC, which means that the documentation pages can be built by Xcode and viewed in the Developer Documentation panel. Build it by clicking Product > Build Documentation
or hitting Shift + Ctrl + Cmd + D
.
To use StatKit, make sure that your system has Swift 5.5 (or later) installed. If you’re using a Mac, also make sure that xcodeselect
points at an Xcode installation that includes a valid version of Swift and that you’re running macOS Catalina (10.15) or later.
IMPORTANT
StatKit does not officially support any beta software, including beta versions of Xcode and macOS, or unreleased versions of Swift.
To install StatKit using the Swift Package Manager, add it as a dependency in your Package.swift
file:
let package = Package(
...
dependencies: [
.package(url: "https://github.com/JimmyMAndersson/StatKit.git", from: "0.5.0")
],
...
)
Then import StatKit where you would like to use it:
import StatKit
StatKit is a young project that is under active development. Our vision is to create the goto statistics library for Swift developers, much like SciPy and NumPy are for the Python language.
❤️ Consider becoming a sponsor to support the development of this library.
You could cover an afternoon coffee or a meal to keep my neurons firing.
Thank you for your contribution, and enjoy using StatKit!