Table of Contents

Time Series Tutorial

Learn how to forecast future values and detect anomalies in temporal data using AiDotNet.

Overview

Time series analysis involves working with data points collected over time. AiDotNet provides 30+ time series models for forecasting, anomaly detection, and pattern recognition.

Available Models

Model Use Case
ARIMA Classical statistical forecasting
SARIMA Seasonal ARIMA
Prophet Facebook's forecasting library
N-BEATS Neural basis expansion
Temporal Fusion Transformer State-of-the-art deep learning
LSTM Long short-term memory networks
GRU Gated recurrent units
Transformer Attention-based forecasting
DeepAR Probabilistic forecasting

Quick Start

using AiDotNet;
using AiDotNet.TimeSeries;

// Sample time series data (monthly sales)
var timestamps = Enumerable.Range(0, 24).Select(i => DateTime.Now.AddMonths(-24 + i)).ToArray();
var values = new double[] { 100, 120, 115, 130, 145, 140, 155, 160, 150, 170, 180, 190,
                            195, 210, 205, 220, 235, 240, 250, 260, 255, 270, 280, 290 };

// Build and train a forecasting model
var result = await new AiModelBuilder<double, double[], double>()
    .ConfigureModel(new LSTMForecaster<double>(
        inputSize: 12,      // Use 12 months of history
        hiddenSize: 64,
        outputSize: 3       // Predict next 3 months
    ))
    .ConfigureOptimizer(new AdamOptimizer<double>())
    .BuildAsync(values);

// Forecast next 3 months
var forecast = result.Predict(values.TakeLast(12).ToArray());
Console.WriteLine($"Forecasted values: {string.Join(", ", forecast)}");

Anomaly Detection

// Detect anomalies in time series
var anomalyDetector = await new AiModelBuilder<double, double[], double>()
    .ConfigureModel(new IsolationForest<double>())
    .BuildAsync(values);

var isAnomaly = anomalyDetector.Predict(newValue);

Next Steps

Edit this page