Doc: Improove docs

Nixtla · Nov 29, 2022 · 1e2bd49 · 1e2bd49
1 parent 1184df3
commit 1e2bd49
Show file tree

Hide file tree

Showing 3 changed files with 150 additions and 56 deletions.
diff --git a/nbs/_quarto.yml b/nbs/_quarto.yml
@@ -10,8 +10,11 @@ format:
     toc: true
 
 website:
-  twitter-card: true
-  open-graph: true
+  twitter-card: 
+    image: "https://farm6.staticflickr.com/5510/14338202952_93595258ff_z.jpg"
+    site: "@Nixtlainc"
+  open-graph:
+    image: "https://github.com/Nixtla/styles/blob/2abf51612584169874c90cd7c4d347e3917eaf73/images/Banner%20Github.png"
   google-analytics: "G-NXJNCVR18L"
   repo-actions: [issue]
   favicon: favicon_png.png

diff --git a/nbs/custom.yml b/nbs/custom.yml
@@ -1,4 +1,5 @@
 website:
+  logo: https://github.com/Nixtla/styles/blob/b9ea432cfa2dae20fc84d8634cae6db902f9ca3f/images/Nixtla_Blanco.png
   reader-mode: false
   navbar:
     collapse-below: lg

diff --git a/nbs/index.ipynb b/nbs/index.ipynb
@@ -7,11 +7,16 @@
    "metadata": {},
    "outputs": [
     {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "The autoreload extension is already loaded. To reload it, use:\n",
-      "  %reload_ext autoreload\n"
+     "ename": "",
+     "evalue": "",
+     "output_type": "error",
+     "traceback": [
+      "\u001b[1;31mJupyter cannot be started. Error attempting to locate Jupyter: Running cells with 'Python 3.9.6 64-bit' requires notebook and jupyter package.\n",
+      "\u001b[1;31mRun the following command to install 'jupyter and notebook' into the Python environment. \n",
+      "\u001b[1;31mCommand: 'python -m pip install jupyter notebook -U\n",
+      "\u001b[1;31mor\n",
+      "\u001b[1;31mconda install jupyter notebook -U'\n",
+      "\u001b[1;31mClick <a href='https://aka.ms/installJupyterForVSCode'>here</a> for more info."
      ]
     }
    ],
@@ -53,32 +58,57 @@
    "source": [
     "## Installation\n",
     "\n",
-    "You can install the *released version* of `StatsForecast` from the [Python package index](https://pypi.org) with:\n",
+    "You can install `StatsForecast` with:\n",
     "\n",
     "```python\n",
     "pip install statsforecast\n",
     "```\n",
-    "We are constantly updating StatsForecast, so we suggest fixing the version to avoid issues. `pip install statsforecast==\"1.0.0\"`\n",
     "\n",
-    "\n",
-    "We recommend installing your libraries inside a python virtual or conda environment. To install the latest *released version* of `StatsForecast` from [conda](https://anaconda.org) run this command:\n",
+    "or \n",
     "\n",
     "```python\n",
     "conda install -c conda-forge statsforecast\n",
-    "```\n",
-    "\n"
+    "``` \n",
+    "\n",
+    "\n",
+    "Vist our [Installation Guide](https://nixtla.github.io/statsforecast/examples/installation.html) for further instructions.\n"
    ]
   },
   {
    "cell_type": "markdown",
    "id": "90f422b7",
    "metadata": {},
    "source": [
-    "## Get Started\n",
+    "## Quick Start\n",
+    "\n",
+    "**Minimal Example**\n",
+    "\n",
+    "```python\n",
+    "from statsforecast import StatsForecast\n",
+    "from statsforecast.models import AutoARIMA\n",
+    "\n",
+    "sf = StatsForecast(\n",
+    "    models = [AutoARIMA(season_length = 12)],\n",
+    "    freq = 'M'\n",
+    ")\n",
+    "\n",
+    "sf.fit(df)\n",
+    "sf.predict(h=12, level=[95])\n",
+    "```\n",
     "\n",
-    "To get started with a minimal example, follow this [guide](./examples/Getting_Started_short.ipynb)\n",
+    "**Get Started with this [quick guide](./examples/Getting_Started_short.ipynb).**\n",
     "\n",
-    "Or follow this [stept by step guide ](./examples/Getting_Started_complete.ipynb) to train, evaluate and select multiple models on many time series. "
+    "**Follow this [end-to-end walkthrough](./examples/Getting_Started_complete.ipynb) for best practices.**"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "2fec4ce5",
+   "metadata": {},
+   "source": [
+    "## Why? \n",
+    "\n",
+    "Current Python alternatives for statistical models are slow, inaccurate and don't scale well. So we created a library that can be used to forecast in production environments or as benchmarks.  `StatsForecast` includes an extensive battery of models that can efficiently fit millions of time series."
    ]
   },
   {
@@ -88,9 +118,10 @@
    "source": [
     "## Features\n",
     "\n",
-    "* Fastest and most accurate implementations of `AutoARIMA`, `ETS`, `AutoCES`, and `MSTL` in Python. \n",
+    "* Fastest and most accurate implementations of `AutoARIMA`, `AutoETS`, `AutoCES`, `MSTL` and `Theta` in Python. \n",
     "* Out-of-the-box compatibility with Spark, Dask, and Ray.\n",
     "* Probabilistic Forecasting and Confidence Intervals.\n",
+    "* Support for exogenous Variables and static covariates.\n",
     "* Anomaly Detection.\n",
     "* Familiar sklearn syntax: `.fit` and `.predict`. \n",
     "* Intermittent demand forecasting with `CrostonClassic`, `CrostonSBA`, `CrostonOptimized`, `ADIDA`, and `IMAPA`.\n",
@@ -106,64 +137,125 @@
     "## Highlights\n",
     "\n",
     "* Inclusion of `exogenous variables` and `prediction intervals` for ARIMA.\n",
-    "* 20x faster than `pmdarima`.\n",
+    "* 20x [faster](./experiments/arima/) than `pmdarima`.\n",
     "* 1.5x faster than `R`.\n",
     "* 500x faster than `Prophet`. \n",
-    "* 4x faster than `statsmodels`.\n",
+    "* 4x [faster](./experiments/ets/) than `statsmodels`.\n",
     "* Compiled to high performance machine code through [`numba`](https://numba.pydata.org/).\n",
     "* 1,000,000 series in [30 min](https://github.com/Nixtla/statsforecast/tree/main/experiments/ray) with [ray](https://github.com/ray-project/ray).\n",
     "* Replace FB-Prophet in two lines of code and gain speed and accuracy. Check the experiments [here](https://github.com/Nixtla/statsforecast/tree/main/experiments/arima_prophet_adapter).\n",
+    "* Fit 10 benchmark models on **1,000,000** series in [under **5 min**](./experiments/benchmarks_at_scale/). \n",
+    "\n",
     "\n",
     "\n",
     "Missing something? Please open an issue or write us in [![Slack](https://img.shields.io/badge/Slack-4A154B?&logo=slack&logoColor=white)](https://join.slack.com/t/nixtlaworkspace/shared_invite/zt-135dssye9-fWTzMpv2WBthq8NK0Yvu6A)\n"
    ]
   },
   {
    "cell_type": "markdown",
-   "id": "a8a9d04c-f064-43d3-8eae-39b69d8afb9e",
+   "id": "25bb6293",
    "metadata": {},
    "source": [
-    "## Why? \n",
+    "## Models\n",
     "\n",
-    "Current Python alternatives for statistical models are slow, inaccurate and don't scale well. So we created a library that can be used to forecast in production environments or as benchmarks.  `StatsForecast` includes an extensive battery of models that can efficiently fit millions of time series."
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "id": "566e6ea0-0174-42a3-94c1-4d1d14591823",
-   "metadata": {},
-   "source": [
-    "## Accuracy & Speed \n",
+    "### Automatic Forecasting\n",
+    "Automatic forecasting tools search for the best parameters and select the best possible model for a series of time series. These tools are useful for large collections of univariate time series.\n",
+    "\n",
+    "|Model | Point Forecast | Probabilistic Forecast | Insample fitted values | Probabilistic fitted values |\n",
+    "|:------|:-------------:|:----------------------:|:---------------------:|:----------------------------:|\n",
+    "|[AutoARIMA](../models.html#autoarima)|✅|✅|✅|✅|✅|\n",
+    "|[AutoETS](../models.html#autoets)|✅|✅|✅|✅|✅|\n",
+    "|[AutoCES](../models.html#autoces)|✅||✅|||\n",
+    "|[AutoTheta](../models.html#autotheta)|✅|✅|✅|✅|✅|\n",
+    ": {tbl-colwidths=\"[75,25]\"}\n",
+    "\n",
+    "### Theta Family\n",
+    "fit two theta lines to a deseasonalized time series, using different techniques to obtain and combine the two theta lines to produce the final forecasts.\n",
+    "\n",
+    "|Model | Point Forecast | Probabilistic Forecast | Insample fitted values | Probabilistic fitted values |\n",
+    "|:------|:-------------:|:----------------------:|:---------------------:|:----------------------------:|\n",
+    "|[Theta](../models.html#theta)|✅|✅|✅|✅|✅|\n",
+    "|[OptimizedTheta](../models.html#optimizedtheta)|✅|✅|✅|✅|✅|\n",
+    "|[DynamicTheta](../models.html#dynamictheta)|✅|✅|✅|✅|✅|\n",
+    "|[DynamicOptimizedTheta](../models.html#dynamicoptimizedtheta)|✅|✅|✅|✅|✅|\n",
+    ": {tbl-colwidths=\"[75,25]\"}\n",
+    "\n",
+    "\n",
+    "### Multiple Seasonalities\n",
+    "Suited for signals with more than one clear seasonality. Useful for low-frequency data like electricity and logs.\n",
     "\n",
-    "### ARIMA \n",
-    "The `AutoARIMA` model implemented in `StatsForecast` is **20x faster** than `pmdarima` and **1.5x faster** than `R`  while improving accuracy. You can see the exact comparison and reproduce the results [here](./experiments/arima/).\n",
+    "|Model | Point Forecast | Probabilistic Forecast | Insample fitted values | Probabilistic fitted values |\n",
+    "|:------|:-------------:|:----------------------:|:---------------------:|:----------------------------:|\n",
+    "|[MSTL](../models.html#mstl)|✅|✅|✅|✅|✅|\n",
+    ": {tbl-colwidths=\"[75,25]\"}\n",
     "\n",
-    "### ETS\n",
+    "### Baseline Models\n",
+    "Classical models for establishing baseline.\n",
     "\n",
-    "StatsForecast's `ETS` is **4x faster than StatsModels'** and **1.6x faster than R's**, with improved accuracy and robustness. You can see the exact comparison and reproduce the results[here](./experiments/ets/)\n",
+    "|Model | Point Forecast | Probabilistic Forecast | Insample fitted values | Probabilistic fitted values |\n",
+    "|:------|:-------------:|:----------------------:|:---------------------:|:----------------------------:|\n",
+    "|[HistoricAverage](../models.html#historicaverage)|✅|✅|✅|✅|✅|\n",
+    "|[Naive](../models.html#naive)|✅|✅|✅|✅|✅|\n",
+    "|[RandomWalkWithDrift](../models.html#randomwalkwithdrift)|✅|✅|✅|✅|✅|\n",
+    "|[SeasonalNaive](../models.html#seasonalnaive)|✅|✅|✅|✅|✅|\n",
+    "|[WindowAverage](../models.html#windowaverage)|✅|||||\n",
+    "|[SeasonalWindowAverage](../models.html#seasonalwindowaverage)|✅|||||\n",
+    ": {tbl-colwidths=\"[75,25]\"}\n",
     "\n",
-    "### Benchmarks at Scale\n",
+    "### Exponential Smoothing\n",
+    "Uses a weighted average of all past observations where the weights decrease exponentially into the past. Suitable for data with no clear trend or seasonality.\n",
     "\n",
-    "With `StatsForecast` you can fit 9 benchmark models on **1,000,000** series in under **5 min**. Reproduce the results [here](./experiments/benchmarks_at_scale/). \n"
+    "|Model | Point Forecast | Probabilistic Forecast | Insample fitted values | Probabilistic fitted values |\n",
+    "|:------|:-------------:|:----------------------:|:---------------------:|:----------------------------:|\n",
+    "|[SimpleExponentialSmoothing](../models.html#simpleexponentialsmoothing)|✅|||||\n",
+    "|[SimpleExponentialSmoothingOptimized](../models.html#simpleexponentialsmoothingoptimized)|✅|||||\n",
+    "|[SeasonalExponentialSmoothing](../models.html#seasonalexponentialsmoothing)|✅|||||\n",
+    "|[SeasonalExponentialSmoothingOptimized](../models.html#seasonalexponentialsmoothingoptimized)|✅|||||\n",
+    "|[Holt](../models.html#holt)|✅|✅|✅|✅|✅|\n",
+    "|[HoltWinters](../models.html#holtwinters)|✅|✅|✅|✅|✅|\n",
+    ": {tbl-colwidths=\"[75,25]\"}\n",
+    "\n",
+    "\n",
+    "\n",
+    "\n",
+    "### Sparse of Inttermitent\n",
+    "Suited for series with very few non-zero observations\n",
+    "\n",
+    "|Model | Point Forecast | Probabilistic Forecast | Insample fitted values | Probabilistic fitted values |\n",
+    "|:------|:-------------:|:----------------------:|:---------------------:|:----------------------------:|\n",
+    "|[ADIDA](../models.html#adida)|✅|||||\n",
+    "|[CrostonClassic](../models.html#crostonclassic)|✅|||||\n",
+    "|[CrostonOptimized](../models.html#crostonoptimized)|✅|||||\n",
+    "|[CrostonSBA](../models.html#crostonsba)|✅|||||\n",
+    "|[IMAPA](../models.html#imapa)|✅|||||\n",
+    "|[TSB](../models.html#tsb)|✅|||||\n",
+    ": {tbl-colwidths=\"[75,25]\"}\n",
+    "\n"
    ]
   },
   {
    "cell_type": "markdown",
-   "id": "8f2477d4-7935-468d-8805-fc674c28461e",
+   "id": "b51b7fbd",
    "metadata": {},
    "source": [
+    "## Examples and Guides\n",
+    "\n",
+    "📚 [End to End Walkthrough](https://nixtla.github.io/statsforecast/examples/getting_started_complete.html): Model training, evaluation and selection for multiple time series\n",
     "\n",
-    "## Getting Started \n",
-    "You can run this notebooks to get you started. \n",
+    "🔎 [Anomaly Detection](https://nixtla.github.io/statsforecast/examples/anomalydetection.html): detect anomalies for time series using in-sample prediction intervals.\n",
     "\n",
-    "* Example of different `AutoARIMA` models on M4 data [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/nixtla/statsforecast/blob/main/nbs/examples/AutoArima_vs_Prophet.ipynb)  \n",
-    "    * In this notebook we present Nixtla's `AutoARIMA`. \n",
-    "    The `AutoARIMA` model is widely used to forecast time series in production and as a benchmark. However, the alternative python implementation (`pmdarima`) is so slow that prevents data scientists from quickly iterating and deploying `AutoARIMA` in production for a large number of time series. \n",
+    "👩‍🔬 [Cross Validation](https://nixtla.github.io/statsforecast/examples/crossvalidation.html): robust model’s performance evaluation.\n",
     "\n",
-    "* Shorter Example of fitting and `AutoARIMA` and an `ETS` model. [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/nixtla/statsforecast/blob/main/nbs/examples/Getting_Started_with_Auto_Arima_and_ETS.ipynb)\n",
+    "❄️ [Multiple Seasonalities](https://nixtla.github.io/statsforecast/examples/multipleseasonalities.html): how to forecast data with multiple seasonalities using an MSTL.\n",
     "\n",
+    "🔌 [Predict Demand Peaks](https://nixtla.github.io/statsforecast/examples/electricitypeakforecasting.html): electricity load forecasting for detecting daily peaks and reducing electric bills.\n",
     "\n",
-    "* Benchmarking 9 models on millions of [series](./experiments/benchmarks_at_scale/)."
+    "📈 [Intermittent Demand](https://nixtla.github.io/statsforecast/examples/intermittentdata.html): forecast series with very few non-zero observations. \n",
+    "\n",
+    "🌡️ [Exogenous Regressors](https://nixtla.github.io/statsforecast/examples/exogenous.html): like weather or prices\n",
+    "\n",
+    "\n",
+    "\n"
    ]
   },
   {
@@ -174,24 +266,22 @@
     "## How to contribute\n",
     "See [CONTRIBUTING.md](https://github.com/Nixtla/statsforecast/blob/main/CONTRIBUTING.md)."
    ]
-  },
-  {
-   "cell_type": "markdown",
-   "id": "880c61b7-ec48-428d-8668-54c9114a51a1",
-   "metadata": {},
-   "source": [
-    "## References\n",
-    "\n",
-    "*  The `AutoARIMA` model is based (translated) from the R implementation included in the [forecast](https://github.com/robjhyndman/forecast) package developed by Rob Hyndman.\n",
-    "*  The `ETS` model is based (translated) from the R implementation included in the [forecast](https://github.com/robjhyndman/forecast) package developed by Rob Hyndman."
-   ]
   }
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "Python 3 (ipykernel)",
+   "display_name": "Python 3.7.12 ('autots-ui')",
    "language": "python",
    "name": "python3"
+  },
+  "language_info": {
+   "name": "python",
+   "version": "3.7.12"
+  },
+  "vscode": {
+   "interpreter": {
+    "hash": "53c8f8a6920a4d601e2858165ee48357c21f0198d63d0b9031b8177ea1f99a56"
+   }
   }
  },
  "nbformat": 4,