Form1.cs

using Microsoft.ML;
using System;
using System.Collections.Generic;
using System.Data;
using System.Drawing;
using System.IO;
using System.Linq;
using System.Windows.Forms;
using System.Windows.Forms.DataVisualization.Charting;

namespace SpikeDetection.WinForms
{
    public partial class Form1 : Form
    {
        private DataTable dataTable = null;
        private string filePath = "";
        Tuple<string, string> tup = null;
        Dictionary<int, Tuple<string, string>> dict = new Dictionary<int, Tuple<string, string>>();
        private static string BaseModelsRelativePath = @"../../../MLModels";
        private static string ModelRelativePath1 = $"{BaseModelsRelativePath}/ProductSalesSpikeModel.zip";
        private static string ModelRelativePath2 = $"{BaseModelsRelativePath}/ProductSalesChangePointModel.zip";


        private static string spikeModelPath = GetAbsolutePath(ModelRelativePath1);
        private static string changePointModelPath = GetAbsolutePath(ModelRelativePath2);

        public Form1()
        {
            InitializeComponent();
        }

        // Find file button.
        private void button1_Click(object sender, EventArgs e)
        {
            // Open File Explorer.
            DialogResult result = openFileExplorer.ShowDialog();

            // Set text in file path textbox to file path from file explorer.
            if (result == DialogResult.OK)
            {
                filePathTextbox.Text = openFileExplorer.FileName;
            }
        }

        // Go button.
        private void button2_Click(object sender, EventArgs e)
        {
            // Set filepath from text from filepath textbox.
            filePath = filePathTextbox.Text;
            
            // Check if file exists.
            if (File.Exists(filePath))
            {
                dict = new Dictionary<int, Tuple<string, string>>();

                if (filePath != "")
                {
                    // Reset text in anomaly textbox.
                    anomalyText.Text = "";

                    // Display preview of dataset and graph.
                    displayDataTableAndGraph();

                    // Load a trained model to detect anomalies and then mark them on the graph.
                    detectAnomalies();

                }
                // If file path textbox is empty, prompt user to input file path.
                else
                {
                    MessageBox.Show("Please input file path.");
                }
            }
            else
            {
                MessageBox.Show("File does not exist. Try finding the file again.");
            }
        }
    


        private void displayDataTableAndGraph()
        {
            dataTable = new DataTable();
            string[] dataCol = null;
            int a = 0;
            string xAxis = "";
            string yAxis = "";

            string[] dataset = File.ReadAllLines(filePath);
            dataCol = commaSeparatedRadio.Checked ? dataset[0].Split(',') : dataset[0].Split('\t');

            dataTable.Columns.Add(dataCol[0]);
            dataTable.Columns.Add(dataCol[1]);
            xAxis = dataCol[0];
            yAxis = dataCol[1];

            foreach (string line in dataset.Skip(1))
            {
                // Add next row of data.
                dataCol = commaSeparatedRadio.Checked ? line.Split(',') : line.Split('\t');
                dataTable.Rows.Add(dataCol);

                tup = new Tuple<string, string>(dataCol[0], dataCol[1]);
                dict.Add(a, tup);

                a++;
            }

            // Set data view preview source.
            dataGridView1.DataSource = dataTable;

            // Update y axis min and max values.
            double yMax = Convert.ToDouble(dataTable.Compute($"max([{yAxis}])", string.Empty));
            //double yMin = Convert.ToDouble(dataTable.Compute($"min([{yAxis}])", string.Empty));
            double yMin = Convert.ToDouble(0);
            // Set graph source.
            graph.DataSource = dataTable;

            // Set graph options.
            graph.Series["Series1"].ChartType = SeriesChartType.Line;

            graph.Series["Series1"].XValueMember = xAxis;
            graph.Series["Series1"].YValueMembers = yAxis;

            graph.Legends["Legend1"].Enabled = true;

            graph.ChartAreas["ChartArea1"].AxisX.MajorGrid.LineWidth = 0;
            graph.ChartAreas["ChartArea1"].AxisX.Interval = a / 10;

            graph.ChartAreas["ChartArea1"].AxisY.Maximum = yMax;
            graph.ChartAreas["ChartArea1"].AxisY.Minimum = yMin;
            graph.ChartAreas["ChartArea1"].AxisY.Interval = yMax / 10;


            graph.DataBind();

        }

        private void detectAnomalies()
        {
            // Create MLContext to be shared across the model creation workflow objects.
            var mlcontext = new MLContext();

            // STEP 1: Load the data into IDataView.
            IDataView dataView = mlcontext.Data.LoadFromTextFile<ProductSalesData>(path: filePath, hasHeader: true, separatorChar: commaSeparatedRadio.Checked ? ',' : '\t');

            // Step 2: Load & use model.
            // Note -- The model is trained with the product-sales dataset in a separate console app (see AnomalyDetectionConsoleApp).
            if (spikeDet.Checked)
            {
                if (File.Exists(spikeModelPath))
                {
                    loadAndUseModel(mlcontext, dataView, spikeModelPath, "Spike", Color.DarkRed);
                }
                else
                {
                    MessageBox.Show("Spike detection model does not exist. Please run model training console app first.");
                }
            }
            if (changePointDet.Checked)
            {

                if (File.Exists(changePointModelPath))
                {
                    loadAndUseModel(mlcontext, dataView, changePointModelPath, "Change point", Color.DarkBlue);
                }
                else
                {
                    MessageBox.Show("Change point detection model does not exist. Please run model training console app first.");
                }
            }
        }

        public static string GetAbsolutePath(string relativePath)
        {
            var _dataRoot = new FileInfo(typeof(Program).Assembly.Location);
            string assemblyFolderPath = _dataRoot.Directory.FullName;

            string fullPath = Path.Combine(assemblyFolderPath, relativePath);

            return fullPath;
        }

        private void loadAndUseModel(MLContext mlcontext, IDataView dataView, String modelPath, String type, Color color)
        {
            ITransformer tansformedModel = mlcontext.Model.Load(modelPath, out var modelInputSchema);
            
            // Step 3: Apply data transformation to create predictions.
            IDataView transformedData = tansformedModel.Transform(dataView);
            var predictions = mlcontext.Data.CreateEnumerable<ProductSalesPrediction>(transformedData, reuseRowObject: false);

            // Index key for dictionary (date, sales).
            int a = 0;

            foreach (var prediction in predictions)
            {
                // Check if anomaly is predicted (indicated by an alert).
                if (prediction.Prediction[0] == 1)
                {
                    // Get the date (year-month) where spike is detected.
                    var xAxisDate = dict[a].Item1;
                    // Get the number of sales which was detected to be a spike.
                    var yAxisSalesNum = dict[a].Item2;

                    // Add anomaly points to graph
                    // and set point/marker options.
                    graph.Series["Series1"].Points[a].SetValueXY(a, yAxisSalesNum);
                    graph.Series["Series1"].Points[a].MarkerStyle = MarkerStyle.Star4;
                    graph.Series["Series1"].Points[a].MarkerSize = 10;
                    graph.Series["Series1"].Points[a].MarkerColor = color;

                    // Print out anomalies as text for user &
                    // change color of text accordingly.
                    string text = type + " detected in " + xAxisDate + ": " + yAxisSalesNum + "\n";
                    anomalyText.SelectionColor = color;
                    anomalyText.AppendText(text);

                    // Change row color in table where anomalies occur.
                    DataGridViewRow row = dataGridView1.Rows[a];
                    row.DefaultCellStyle.BackColor = color;
                    row.DefaultCellStyle.ForeColor = Color.White;
                }
                a++;
            }
        }

        private void Form1_Load(object sender, EventArgs e)
        {

        }

        private void RadioButton1_CheckedChanged(object sender, EventArgs e)
        {

        }
    }
}