Turkish Names Gender Prediction

GenderPrediction.Turkish is a library that provides prediction of gender from Turkish names built on ML.NET Both training application and library included in source code.

Supported Platforms

• .NET 4.6.1 (Desktop / Server)
• .NET Standard 2.0
• .NET Core 2.0, 2.1, 2.2

Continuous integration

Build server	Platform	Build status	Integration tests
AppVeyor	Windows
Travis	Linux / MacOS

Table of Contents

1. Installation
2. Usage
   • Standalone Initialization
   • Microsoft.Extensions.DependencyInjection Initialization
3. License

Installation

Following commands can be used to install GenderPrediction.Turkish, run the following command in the Package Manager Console

Install-Package GenderPrediction.Turkish

Or use dotnet cli

dotnet GenderPrediction.Turkish

Usage

GenderPrediction.Turkish can be used with any DI library, or it can be used standalone.

Standalone Initialization

If you do not want to use any DI framework, you have to instantiate GenderPredictionStandalone as follows.

IGenderPredictionService genderPredictionService = GenderPredictionStandalone.Create();

Microsoft.Extensions.DependencyInjection Initialization

First, you need to install Microsoft.Extensions.DependencyInjection NuGet package as follows

dotnet add package Microsoft.Extensions.DependencyInjection

Register necessary dependencies to ServiceCollection as follows

var services = new ServiceCollection();
services.AddSingleton<IGenderPredictionEngine, GenderPredictionEngine>();
services.AddTransient<IGenderPredictionService, GenderPredictionService>();

ServiceProvider buildServiceProvider = services.BuildServiceProvider();

var genderPredictionService = buildServiceProvider.GetRequiredService<IGenderPredictionService>();

The singleton registration of the GenderPredictionEngine is important for performance, since its loads and initialize the trained model. After the library is initialized, it is very easy to use.

GenderPredictionModel model = genderPredictionService.Predict("Deniz");

string name = genderPredictionModel.Name;
Gender predictedGender = genderPredictionModel.PredictedGender;
float maleProbability = genderPredictionModel.Score[Gender.Male];
float femaleProbability = genderPredictionModel.Score[Gender.Female];
float unisexProbability = genderPredictionModel.UnisexProbability

Or

string[] names = new[] {"Dilek", "Hasan", "Mehmet", "İbrahim"};
IEnumerable<GenderPredictionModel> model = genderPredictionService.Predict(names);

License

Licensed under MIT, see LICENSE for the full text.