FEAT: online anomaly detection (#546)

Co-authored-by: yibeihu <yibei2022@163.com>
Co-authored-by: Yibei Hu <115324670+Yibei990826@users.noreply.github.com>

nbs/docs/capabilities/historical-anomaly-detection/00_historical_anomaly_detection.ipynb

@@ -0,0 +1,43 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "6de758ee-a0d2-4b3f-acff-eed419dd17c5",
+   "metadata": {},
+   "source": [
+    "# Historical anomaly detection"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "5d267032-535b-4b7b-b7d3-d2db8f673af6",
+   "metadata": {},
+   "source": [
+    "This section provides various recipes for performing historical anomaly detection using TimeGPT.\n",
+    "\n",
+    "Historical anomaly detection identifies data points that deviate from the expected behavior over a given historical time series, helping to spot fraudulent activity, security breaches, or significant outliers.\n",
+    "\n",
+    "The process involves generating predictions and constructing a 99% confidence interval. Data points falling outside this interval are considered anomalies.\n",
+    "\n",
+    "This section covers:\n",
+    "\n",
+    "* [Historical anomaly detection](https://docs.nixtla.io/docs/capabilities-historical-anomaly-detection-quickstart)\n",
+    "\n",
+    "* [Historical anomaly detection with exogenous features](https://docs.nixtla.io/docs/capabilities-historical-anomaly-detection-add_exogenous_variables)\n",
+    "\n",
+    "* [Historical anomaly detection with date features](https://docs.nixtla.io/docs/capabilities-historical-anomaly-detection-add_date_features)\n",
+    "\n",
+    "* [Modifying the confidence intervals](https://docs.nixtla.io/docs/capabilities-historical-anomaly-detection-add_confidence_levels)"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "python3",
+   "language": "python",
+   "name": "python3"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

nbs/docs/capabilities/anomaly-detection/01_quickstart.ipynb → nbs/docs/capabilities/historical-anomaly-detection/01_quickstart.ipynb

@@ -48,7 +48,7 @@
    "source": [
     "# Quickstart\n",
     "\n",
-    "To perform anomaly detection, use the `detect_anomalies` method. Then, plot the anomalies using the `plot` method."
+    "To perform historical anomaly detection, use the `detect_anomalies` method. Then, plot the anomalies using the `plot` method."
    ]
   },
   {
@@ -73,7 +73,7 @@
     "#| echo: false\n",
     "if not IN_COLAB:\n",
     "    load_dotenv()\n",
-    "    colab_badge('docs/capabilities/anomaly-detection/01_quickstart')"
+    "    colab_badge('docs/capabilities/historical-anomaly-detection/01_quickstart')"
    ]
   },
   {
@@ -176,7 +176,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "For an in-depth guide on anomaly detection with TimeGPT, check out our [tutorial](https://docs.nixtla.io/docs/tutorials-anomaly_detection)."
+    "For an in-depth guide on historical anomaly detection with TimeGPT, check out our [tutorial](https://docs.nixtla.io/docs/tutorials-anomaly_detection)."
    ]
   }
  ],

nbs/docs/capabilities/anomaly-detection/04_confidence_levels.ipynb → nbs/docs/capabilities/historical-anomaly-detection/04_confidence_levels.ipynb

@@ -48,7 +48,7 @@
    "source": [
     "# Add confidence levels\n",
     "\n",
-    "Tweak the confidence level used for anomaly detection. By default, if a value falls outside the 99% confidence interval, it is labeled as an anomaly.\n",
+    "Tweak the confidence level used for historical anomaly detection. By default, if a value falls outside the 99% confidence interval, it is labeled as an anomaly.\n",
     "\n",
     "Modify this with the `level` parameter, which accepts any value between 0 and 100, including decimals.\n",
     "\n",
@@ -77,7 +77,7 @@
     "#| echo: false\n",
     "if not IN_COLAB:\n",
     "    load_dotenv()\n",
-    "    colab_badge('docs/capabilities/anomaly-detection/04_confidence_levels')"
+    "    colab_badge('docs/capabilities/historical-anomaly-detection/04_confidence_levels')"
    ]
   },
   {

nbs/docs/capabilities/online-anomaly-detection/00_online_anomaly_detection.ipynb

@@ -0,0 +1,36 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Online (Real-Time) Anomaly Detection"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Online anomaly detection dynamically identifies anomalies as data streams in, allowing users to specify the number of timestamps to monitor. This method is well-suited for immediate applications, such as fraud detection, live sensor monitoring, or tracking real-time demand changes. By focusing on recent data and continuously generating forecasts, it enables timely responses to anomalies in critical scenarios.\n",
+    "\n",
+    "This section provides various recipes for performing real-time anomaly detection using TimeGPT, offering users the ability to detect outliers and unusual patterns as they emerge, ensuring prompt intervention in time-sensitive situations.\n",
+    "\n",
+    "This section covers:\n",
+    "\n",
+    "* [Online anomaly detection](https://docs.nixtla.io/docs/capabilities-online-anomaly-detection-quickstart)\n",
+    "\n",
+    "* [How to adjust the detection process](https://docs.nixtla.io/docs/capabilities-online-anomaly-detection-adjusting_detection_process.ipynb)\n",
+    "\n",
+    "* [Univariate vs. multiseries anomaly detection](https://docs.nixtla.io/docs/capabilities-online-anomaly-detection-univariate_vs_multivariate_anomaly_detection)\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": []
+  }
+ ],
+ "metadata": {},
+ "nbformat": 4,
+ "nbformat_minor": 2
+}