Maptor

این صفحه هنوز ترجمه نشده است؛ نسخهٔ انگلیسی نمایش داده می‌شود.

IRI.Maptor.Core.MachineLearning

NuGet Target

Machine learning algorithms and statistical tools used in the Maptor stack, with a focus on spatial data analysis — clustering, association rule mining, logistic regression, and descriptive statistics.

Installation

dotnet add package IRI.Maptor.Core.MachineLearning

Features

  • DBSCAN density-based clustering (Dbscan.Cluster<T>) with a caller-supplied distance function
  • Apriori frequent-itemset mining (AprioriAlgorithm, Itemset)
  • Binary logistic regression (LogisticRegression) with configurable options: feature normalization, regularization method, variance calculation mode
  • Logistic-regression-based line simplification use case (LogisticSimplification with synthetic training data helpers)
  • Descriptive statistics (GeneralStatistics.CalculateSummary)
  • Shared building blocks: sigmoid, normalization, and regularization helpers

Usage

Train and use a logistic regression classifier:

using IRI.Maptor.Core.MachineLearning;
using IRI.Maptor.Core.Common.Mathematics;

var options = new LogisticRegressionOptions { NormalizeFeatures = true };
var lr = new LogisticRegression(options);

// each row of xValues is one observation; yValues holds the 0/1 labels
lr.Fit(xValues, yValues);   // xValues: Matrix, yValues: double[]

double? probability = lr.Predict(new List<double> { 2.5, 1.3 });

Summarize a data series:

using IRI.Maptor.Core.MachineLearning;

var summary = GeneralStatistics.CalculateSummary(new double[] { 1, 2, 3, 4, 5 });

Limitations

  • Dbscan.Cluster currently returns void; the computed cluster assignments are not yet exposed to the caller.
  • Logistic regression is binary-only (no multi-class support).

NuGet package · Report issues · Back to IRI.Maptor.Core