qlhb/Ropin.IOT.MLService/Program.cs

using Microsoft.AspNetCore.Hosting;
using Microsoft.Extensions.Configuration;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;
using System;
using System.Collections.Generic;
using System.Linq;
using System.Threading.Tasks;
using Microsoft.ML;
using Microsoft.ML.Data;

namespace Ropin.IOT.MLService
{
    public class Program
    {
        // 定义文件路径
        private static string _dataPath = "./accident_data.csv";
        private static string _modelPath = "./AccidentPredictionModel.zip";
        public static void Main(string[] args)
        {
            
            Console.OutputEncoding = System.Text.Encoding.UTF8;
            Console.WriteLine("工厂事故预测系统启动...");

            // 创建ML.NET上下文  
            MLContext mlContext = new MLContext(seed: 0);

            // 加载数据  
            Console.WriteLine("正在加载数据...");
            IDataView dataView = mlContext.Data.LoadFromTextFile<AccidentData>(
                path: _dataPath,
                hasHeader: true,
                separatorChar: ',',
                allowQuoting: true
            );

            // 分割训练集和测试集  
            var dataSplit = mlContext.Data.TrainTestSplit(dataView, testFraction: 0.2);
            var trainingData = dataSplit.TrainSet;
            var testData = dataSplit.TestSet;

            // 数据处理和特征工程  
            Console.WriteLine("正在处理数据和提取特征...");
            var dataProcessPipeline = mlContext.Transforms.Conversion.MapValueToKey(
                    outputColumnName: "Label",  // 这将创建一个新列作为Key类型标签  
                    inputColumnName: "AccidentLevel"  // 原始字符串标签  
                )
                .Append(mlContext.Transforms.Categorical.OneHotEncoding(
                    new[] {
                new InputOutputColumnPair("CountryEncoded", "Country"),
                new InputOutputColumnPair("IndustrySectorEncoded", "IndustrySector"),
                new InputOutputColumnPair("CriticalRiskEncoded", "CriticalRisk"),
                new InputOutputColumnPair("GenreEncoded", "Genre"),
                new InputOutputColumnPair("EmployeeTypeEncoded", "EmployeeOrThirdParty")
                    }))
                // 文本特征提取（从描述中提取特征）  
                .Append(mlContext.Transforms.Text.FeaturizeText("DescriptionFeaturized", "Description"))
                // 合并所有特征到一个向量  
                .Append(mlContext.Transforms.Concatenate("Features",
                    "CountryEncoded", "IndustrySectorEncoded", "CriticalRiskEncoded",
                    "GenreEncoded", "EmployeeTypeEncoded", "DescriptionFeaturized"));

            // 选择训练算法 - 使用多类分类器  
            Console.WriteLine("构建和训练模型...");
            var trainer = mlContext.MulticlassClassification.Trainers.SdcaMaximumEntropy();

            // 构建完整训练管道  
            var trainingPipeline = dataProcessPipeline.Append(trainer)
                // 将预测结果映射回原始标签值  
                .Append(mlContext.Transforms.Conversion.MapKeyToValue("PredictedLabel"));

            // 训练模型  
            var trainedModel = trainingPipeline.Fit(trainingData);
            Console.WriteLine("模型训练完成！");

            // 保存模型  
            mlContext.Model.Save(trainedModel, trainingData.Schema, _modelPath);
            Console.WriteLine($"模型已保存到: {_modelPath}");

            // 评估模型  
            Console.WriteLine("评估模型性能...");
            var predictions = trainedModel.Transform(testData);
            var metrics = mlContext.MulticlassClassification.Evaluate(predictions);

            // 输出评估指标  
            Console.WriteLine($"宏平均精确度: {metrics.MacroAccuracy:F2}");
            Console.WriteLine($"微平均精确度: {metrics.MicroAccuracy:F2}");
            Console.WriteLine($"对数损失: {metrics.LogLoss:F2}");
            Console.WriteLine($"混淆矩阵: \n{metrics.ConfusionMatrix.GetFormattedConfusionTable()}");

            // 使用模型进行预测示例  
            PredictSample(mlContext, trainedModel);
            CreateHostBuilder(args).Build().Run();
        }

        public static IHostBuilder CreateHostBuilder(string[] args) =>
            Host.CreateDefaultBuilder(args)
                .ConfigureWebHostDefaults(webBuilder =>
                {
                    webBuilder.UseStartup<Startup>();
                });
        private static void PredictSample(MLContext mlContext, ITransformer model)
        {
            // 创建预测引擎  
            var predictionEngine = mlContext.Model.CreatePredictionEngine<AccidentData, AccidentPrediction>(model);

            // 创建测试样本  
            var sampleAccident = new AccidentData
            {
                Country = "Country_01",
                IndustrySector = "Mining",
                Genre = "Male",
                EmployeeOrThirdParty = "Third Party",
                CriticalRisk = "Pressed",
                Description = "Worker operating drilling equipment without proper safety procedures, potential for hand injury."
            };

            // 进行预测  
            var prediction = predictionEngine.Predict(sampleAccident);

            Console.WriteLine("\n预测示例：");
            Console.WriteLine($"行业: {sampleAccident.IndustrySector}");
            Console.WriteLine($"风险类型: {sampleAccident.CriticalRisk}");
            Console.WriteLine($"事故描述: {sampleAccident.Description}");
            Console.WriteLine($"预测事故等级: {prediction.PredictedAccidentLevel}");

            // 输出各类别的概率  
            Console.WriteLine("各等级的概率分布:");
            var labels = new[] { "I", "II", "III", "IV" };
            for (int i = 0; i < prediction.Score.Length; i++)
            {
                if (i < prediction.Score.Length)
                    Console.WriteLine($"等级 {i}: {prediction.Score[i]:P2}");
            }

            Console.ReadKey();
        }
    }
    public class AccidentData
    {
        [LoadColumn(0)]
        public float SeqNo { get; set; }

        [LoadColumn(1)]
        public string Date { get; set; }

        [LoadColumn(2)]
        public string Country { get; set; }

        [LoadColumn(3)]
        public string Local { get; set; }

        [LoadColumn(4)]
        public string IndustrySector { get; set; }

        [LoadColumn(5)]
        public string AccidentLevel { get; set; }  // 原始字符串标签  

        [LoadColumn(6)]
        public string PotentialAccidentLevel { get; set; }

        [LoadColumn(7)]
        public string Genre { get; set; }

        [LoadColumn(8)]
        public string EmployeeOrThirdParty { get; set; }

        [LoadColumn(9)]
        public string CriticalRisk { get; set; }

        [LoadColumn(10)]
        public string Description { get; set; }
    }

    public class AccidentPrediction
    {
        [ColumnName("PredictedLabel")]
        public string PredictedAccidentLevel { get; set; }

        public float[] Score { get; set; }
    }
}