{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Standardized Research\n",
    "\n",
    "This tutorial covers in a simple example of a Lasso regression model all steps in the development process of a new \n",
    "model that follows the standards of NinoLearn.\n",
    "\n",
    "## Download\n",
    "\n",
    "Download four indeces which we want to use to predict the Oceaninc Nino Index (ONI):\n",
    "\n",
    "1. The ONI index itself.\n",
    "2. The Dipole Mode Index of the Indian Ocean Dipole (IOD)\n",
    "3. The Warm Water Volume (WWV)\n",
    "4. The Kiritimati Index that can be used as WWV proxy form 1955 onwards.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "oni.txt already downloaded\n",
      "iod.txt already downloaded\n",
      "wwv.dat already downloaded\n",
      "Copy Kindex.mat to data directory\n"
     ]
    }
   ],
   "source": [
    "from ninolearn.download import download, sources\n",
    "\n",
    "download(sources.ONI)\n",
    "download(sources.IOD)\n",
    "download(sources.WWV)\n",
    "download(sources.KINDEX)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Prepare data\n",
    "\n",
    "Extract the essential data form the raw files and move them into preprocessed data directory. If you are interested what these functions exactly do, check out there source code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Prepare ONI timeseries.\n",
      "Prepare WWV  timeseries.\n",
      "Prepare IOD timeseries.\n"
     ]
    }
   ],
   "source": [
    "from ninolearn.preprocess.prepare import prep_oni, prep_wwv\n",
    "from ninolearn.preprocess.prepare import prep_iod, prep_K_index, prep_wwv_proxy\n",
    "\n",
    "\n",
    "prep_oni()\n",
    "prep_wwv()\n",
    "prep_iod()\n",
    "prep_K_index()\n",
    "\n",
    "# combines the WWV and the K-Index to one WWV proxy\n",
    "prep_wwv_proxy()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Build a new Model\n",
    "\n",
    "Here an example of an multilinear (Lasso) regression model is given that is based on the scikit-learn python package.\n",
    "\n",
    "### Data pipeline\n",
    "First a data pipeline is build. The pipeline is used during training, prediction and evaluation to generate the feature, the label, the time as well as (optional) the persistance forecast.\n",
    "\n",
    "When you build a new data pipeline, it needs to have the same structure as the code block below. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import the data reader to read data from the preprocessed data directory\n",
    "from ninolearn.IO.read_processed import data_reader\n",
    "import numpy as np\n",
    "\n",
    "def pipeline(lead_time,  return_persistance=False):\n",
    "    \"\"\"\n",
    "    Data pipeline for the processing of the data before the lasso regression model\n",
    "    is trained.\n",
    "\n",
    "    :type lead_time: int\n",
    "    :param lead_time: The lead time in month.\n",
    "\n",
    "    :type return_persistance: boolean\n",
    "    :param return_persistance: Return as the persistance as well.\n",
    "\n",
    "    :returns: The feature \"X\" (at observation time), the label \"y\" (at lead\n",
    "    time), the target season \"timey\" (least month) and if selected the\n",
    "    label at observation time \"y_persistance\". Hence, the output comes as:\n",
    "    X, y, timey, y_persistance.\n",
    "    \"\"\"\n",
    "    # initialize the reader\n",
    "    reader = data_reader(startdate='1960-01', enddate='2017-12')\n",
    "\n",
    "    # Load data \n",
    "    # HERE you could load other data sources\n",
    "    oni = reader.read_csv('oni')\n",
    "    wwv = reader.read_csv('wwv_proxy')\n",
    "    iod = reader.read_csv('iod')\n",
    "    \n",
    "    # the shift data by 3 in addition to lead time shift (due to definition\n",
    "    # of lead time) as in barnston et al. (2012)\n",
    "    shift = 3\n",
    "    \n",
    "    # Make feature\n",
    "    # HERE you need to stack you data if you loaded different data sets\n",
    "    Xorg = np.stack((oni, wwv, iod), axis=1)\n",
    "    X = Xorg[:-lead_time-shift,:]\n",
    "\n",
    "    # arange label\n",
    "    yorg = oni.values\n",
    "    y = yorg[lead_time + shift:]\n",
    "\n",
    "    # get the time axis of the label\n",
    "    timey = oni.index[lead_time + shift:]\n",
    "\n",
    "    if return_persistance:\n",
    "        y_persistance = yorg[: - lead_time - shift]\n",
    "        return X, y, timey, y_persistance\n",
    "    else:\n",
    "        return X, y, timey"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### The model\n",
    "\n",
    "First of all, the newly designed model needs to inherit from the baseModel such that the model can be trained and evaluated in a standardized procedure furhter down.\n",
    "\n",
    "The model has mandatory variables and functions that need to be included. These parts are highlighted in the following with the comment \"MANDATORY\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "# import the baseModel from which the mlr class needs to inherit\n",
    "from ninolearn.learn.models.baseModel import baseModel\n",
    "\n",
    "# import the sklearn model that we want to use for the ENSO forecast\n",
    "from sklearn.linear_model import Lasso\n",
    "\n",
    "# import some packages and methods to saving the model later\n",
    "import pickle\n",
    "from os.path import join, exists\n",
    "from os import mkdir\n",
    "\n",
    "# MANDATORY: Needs to in herit from the class baseModel\n",
    "class mlr(baseModel):\n",
    "\n",
    "    # MANDETORY: Define how many outputs your model has\n",
    "    n_outputs=1\n",
    "    \n",
    "    # MANDETORY: The name that is used when predictions are saved in an netCDF file.\n",
    "    output_names = ['prediction']\n",
    "\n",
    "    \n",
    "    # MANDETORY: The model needs to have a .__init__() method.\n",
    "    def __init__(self, alpha=1.0, name='mlr'):\n",
    "        \"\"\"\n",
    "        The model needs to have an __init__ function. That takes contains\n",
    "        receives the hyperparameters of the model as well as the name of the\n",
    "        model as keyword arguments\n",
    "        \n",
    "        :type alpha: float\n",
    "        :param alpha: The coefficent for the lasso penatly term.\n",
    "        \n",
    "        :type name: str\n",
    "        :param name: The name of the model that is used to save it to a file after training.\n",
    "        \"\"\"\n",
    "        #  MANDETORY: Apply the .set_hyperparameters function to all keyword arguments.\n",
    "        self.set_hyperparameters(alpha=alpha, name=name)\n",
    "\n",
    "        \n",
    "        \n",
    "    # MANDETORY: The model needs to have a .fit() function that takes trainX, trainy as arguments. \n",
    "    # Very complex models, e.g. neural networks would need to split the trainX and trainy variables further\n",
    "    # to generate a validation data set, which is than used to calculate\n",
    "    # the self.mean_val_loss and to check for overfitting.\n",
    "    # Here, we don't need to do so because the model is not very complex and\n",
    "    # we have plenty of data to train the model.\n",
    "    def fit(self, trainX, trainy):\n",
    "        \"\"\"\n",
    "        This is the fit function of the model. \n",
    "        \n",
    "        :param trainX: The features.\n",
    "        :param trainy: The label.\n",
    "        \"\"\"\n",
    "        #Initialize the Lasso model form the sci-kit learn package\n",
    "        self.model = Lasso(self.hyperparameters['alpha'])\n",
    "\n",
    "        # fit the model to the training data\n",
    "        self.model.fit(trainX,trainy)\n",
    "\n",
    "        # MANDETORY: Save the Score under self.mean_val_loss. This variable\n",
    "        # will be used to be optimized during the random search later\n",
    "        self.mean_val_loss = self.model.score(trainX, trainy)\n",
    "    \n",
    "    # MANDATORY: The model needs to have a .fit() function that takes as arguments. \n",
    "    def predict(self, X):\n",
    "        \"\"\"\n",
    "        Make a prediction.\n",
    "        \n",
    "        :param: A feature set.\n",
    "        \"\"\"\n",
    "        # MANDATORY: Function needs to return a value (the prediction).\n",
    "        return self.model.predict(X)\n",
    "\n",
    "    # MANDATORY: The model needs to have a .save() function. The location where to save the model \n",
    "    # is defined by the keyword arguments 'location' and 'name'\n",
    "    def save(self, location='', dir_name='mlr'):\n",
    "        \"\"\"\n",
    "        Arguments of this function are mandetory and used to systemically\n",
    "        save models in your modeldir.\n",
    "        \"\"\"\n",
    "        path = join(location, dir_name)\n",
    "        if not exists(path):\n",
    "            mkdir(path)\n",
    "        filename = join(path,f'model.sav')\n",
    "        pickle.dump(self.model, open(filename, 'wb'))\n",
    "    \n",
    "    # MANDATORY: The model needs to have a .load() function. The location where to saved model can be found \n",
    "    # is defined by the keyword arguments 'location' and 'name'\n",
    "    def load(self, location='', dir_name='mlr'):\n",
    "        \"\"\"\n",
    "        Arguments of this function are mandetory and used to systemically\n",
    "        load models from your modeldir.\n",
    "        \"\"\"\n",
    "        path = join(location, dir_name)\n",
    "        filename = join(path,f'model.sav')\n",
    "        self.model = pickle.load(open(filename, 'rb'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Cross train the model\n",
    "\n",
    "In the cross_training() function the model is trained on a 5 of 6 time \"decades\" (1962-1971, 1972-1981,..., 2012-2018). For each decade, 50 search iterations with a random uniform choice of `alpha` between 0. and 0.001 is performed. The model that has the best score (in terms of `self.mean_val_loss`, see above) is saved in the model directory."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "##################################################################\n",
      "Lead time: 0 month\n",
      "##################################################################\n",
      "\n",
      "Test period: 1963-01-01 till 1971-12-01\n",
      "--------------------------------------\n",
      "Search iteration Nr 1/10\n",
      "New best hyperparameters\n",
      "Mean loss: 0.8094758951404967\n",
      "{'alpha': 0.0006389865108761426, 'name': 'mlr'}\n",
      "Search iteration Nr 2/10\n",
      "New best hyperparameters\n",
      "Mean loss: 0.8094692817259667\n",
      "{'alpha': 0.0009598393037831796, 'name': 'mlr'}\n",
      "Search iteration Nr 3/10\n",
      "Search iteration Nr 4/10\n",
      "Search iteration Nr 5/10\n",
      "Search iteration Nr 6/10\n",
      "Search iteration Nr 7/10\n",
      "Search iteration Nr 8/10\n",
      "Search iteration Nr 9/10\n",
      "Search iteration Nr 10/10\n",
      "New best hyperparameters\n",
      "Mean loss: 0.8094686842331568\n",
      "{'alpha': 0.0009836857556539448, 'name': 'mlr'}\n",
      "Refit the model with best hyperparamters\n",
      "{'alpha': 0.0009836857556539448, 'name': 'mlr'}\n",
      "best loss search: 0.8094686842331568\n",
      "loss refitting : 0.8094686842331568\n",
      "mlr_decade1972_lead0 already exists\n",
      "mlr_decade1982_lead0 already exists\n",
      "mlr_decade1992_lead0 already exists\n",
      "mlr_decade2002_lead0 already exists\n",
      "mlr_decade2012_lead0 already exists\n",
      "\n",
      "##################################################################\n",
      "Lead time: 3 month\n",
      "##################################################################\n",
      "\n",
      "Test period: 1963-01-01 till 1971-12-01\n",
      "--------------------------------------\n",
      "Search iteration Nr 1/10\n",
      "New best hyperparameters\n",
      "Mean loss: 0.4886563086165414\n",
      "{'alpha': 0.000606099677219931, 'name': 'mlr'}\n",
      "Search iteration Nr 2/10\n",
      "Search iteration Nr 3/10\n",
      "Search iteration Nr 4/10\n",
      "New best hyperparameters\n",
      "Mean loss: 0.48865291157854074\n",
      "{'alpha': 0.0007944042759574363, 'name': 'mlr'}\n",
      "Search iteration Nr 5/10\n",
      "Search iteration Nr 6/10\n",
      "Search iteration Nr 7/10\n",
      "Search iteration Nr 8/10\n",
      "Search iteration Nr 9/10\n",
      "Search iteration Nr 10/10\n",
      "Refit the model with best hyperparamters\n",
      "{'alpha': 0.0007944042759574363, 'name': 'mlr'}\n",
      "best loss search: 0.48865291157854074\n",
      "loss refitting : 0.48865291157854074\n",
      "mlr_decade1972_lead3 already exists\n",
      "mlr_decade1982_lead3 already exists\n",
      "mlr_decade1992_lead3 already exists\n",
      "mlr_decade2002_lead3 already exists\n",
      "mlr_decade2012_lead3 already exists\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/paul/miniconda2/envs/ninolearn/lib/python3.6/site-packages/sklearn/linear_model/coordinate_descent.py:475: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations. Duality gap: 107.17116734450472, tolerance: 0.042875161649484544\n",
      "  positive)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "##################################################################\n",
      "Lead time: 6 month\n",
      "##################################################################\n",
      "\n",
      "Test period: 1963-01-01 till 1971-12-01\n",
      "--------------------------------------\n",
      "Search iteration Nr 1/10\n",
      "New best hyperparameters\n",
      "Mean loss: 0.27359880477009446\n",
      "{'alpha': 0.0005106845288966993, 'name': 'mlr'}\n",
      "Search iteration Nr 2/10\n",
      "New best hyperparameters\n",
      "Mean loss: 0.27359709002505495\n",
      "{'alpha': 0.000626995025067661, 'name': 'mlr'}\n",
      "Search iteration Nr 3/10\n",
      "New best hyperparameters\n",
      "Mean loss: 0.2735936975314077\n",
      "{'alpha': 0.0008092687773921267, 'name': 'mlr'}\n",
      "Search iteration Nr 4/10\n",
      "Search iteration Nr 5/10\n",
      "Search iteration Nr 6/10\n",
      "New best hyperparameters\n",
      "Mean loss: 0.27359327141677314\n",
      "{'alpha': 0.0008293361949688865, 'name': 'mlr'}\n",
      "Search iteration Nr 7/10\n",
      "Search iteration Nr 8/10\n",
      "Search iteration Nr 9/10\n",
      "Search iteration Nr 10/10\n",
      "New best hyperparameters\n",
      "Mean loss: 0.2735902578947911\n",
      "{'alpha': 0.0009593472221216335, 'name': 'mlr'}\n",
      "Refit the model with best hyperparamters\n",
      "{'alpha': 0.0009593472221216335, 'name': 'mlr'}\n",
      "best loss search: 0.2735902578947911\n",
      "loss refitting : 0.2735902578947911\n",
      "mlr_decade1972_lead6 already exists\n",
      "mlr_decade1982_lead6 already exists\n",
      "mlr_decade1992_lead6 already exists\n",
      "mlr_decade2002_lead6 already exists\n",
      "mlr_decade2012_lead6 already exists\n",
      "\n",
      "##################################################################\n",
      "Lead time: 9 month\n",
      "##################################################################\n",
      "\n",
      "Test period: 1963-01-01 till 1971-12-01\n",
      "--------------------------------------\n",
      "Search iteration Nr 1/10\n",
      "New best hyperparameters\n",
      "Mean loss: 0.20925019869561215\n",
      "{'alpha': 0.0006824009115007106, 'name': 'mlr'}\n",
      "Search iteration Nr 2/10\n",
      "Search iteration Nr 3/10\n",
      "Search iteration Nr 4/10\n",
      "New best hyperparameters\n",
      "Mean loss: 0.20924630487479634\n",
      "{'alpha': 0.000874571154763512, 'name': 'mlr'}\n",
      "Search iteration Nr 5/10\n",
      "Search iteration Nr 6/10\n",
      "New best hyperparameters\n",
      "Mean loss: 0.20924357113764858\n",
      "{'alpha': 0.0009873888944211854, 'name': 'mlr'}\n",
      "Search iteration Nr 7/10\n",
      "Search iteration Nr 8/10\n",
      "Search iteration Nr 9/10\n",
      "Search iteration Nr 10/10\n",
      "Refit the model with best hyperparamters\n",
      "{'alpha': 0.0009873888944211854, 'name': 'mlr'}\n",
      "best loss search: 0.20924357113764858\n",
      "loss refitting : 0.20924357113764858\n",
      "mlr_decade1972_lead9 already exists\n",
      "mlr_decade1982_lead9 already exists\n",
      "mlr_decade1992_lead9 already exists\n",
      "mlr_decade2002_lead9 already exists\n",
      "mlr_decade2012_lead9 already exists\n",
      "\n",
      "##################################################################\n",
      "Lead time: 12 month\n",
      "##################################################################\n",
      "\n",
      "Test period: 1963-01-01 till 1971-12-01\n",
      "--------------------------------------\n",
      "Search iteration Nr 1/10\n",
      "New best hyperparameters\n",
      "Mean loss: 0.20681268398908717\n",
      "{'alpha': 0.0009776809179163517, 'name': 'mlr'}\n",
      "Search iteration Nr 2/10\n",
      "Search iteration Nr 3/10\n",
      "Search iteration Nr 4/10\n",
      "Search iteration Nr 5/10\n",
      "Search iteration Nr 6/10\n",
      "Search iteration Nr 7/10\n",
      "Search iteration Nr 8/10\n",
      "Search iteration Nr 9/10\n",
      "Search iteration Nr 10/10\n",
      "Refit the model with best hyperparamters\n",
      "{'alpha': 0.0009776809179163517, 'name': 'mlr'}\n",
      "best loss search: 0.20681268398908717\n",
      "loss refitting : 0.20681268398908717\n",
      "mlr_decade1972_lead12 already exists\n",
      "mlr_decade1982_lead12 already exists\n",
      "mlr_decade1992_lead12 already exists\n",
      "mlr_decade2002_lead12 already exists\n",
      "mlr_decade2012_lead12 already exists\n",
      "\n",
      "##################################################################\n",
      "Lead time: 15 month\n",
      "##################################################################\n",
      "\n",
      "Test period: 1963-01-01 till 1971-12-01\n",
      "--------------------------------------\n",
      "Search iteration Nr 1/10\n",
      "New best hyperparameters\n",
      "Mean loss: 0.1917244585769392\n",
      "{'alpha': 0.0005871325405723177, 'name': 'mlr'}\n",
      "Search iteration Nr 2/10\n",
      "New best hyperparameters\n",
      "Mean loss: 0.19172079586220403\n",
      "{'alpha': 0.0007872452638641191, 'name': 'mlr'}\n",
      "Search iteration Nr 3/10\n",
      "Search iteration Nr 4/10\n",
      "Search iteration Nr 5/10\n",
      "Search iteration Nr 6/10\n",
      "Search iteration Nr 7/10\n",
      "Search iteration Nr 8/10\n",
      "Search iteration Nr 9/10\n",
      "Search iteration Nr 10/10\n",
      "Refit the model with best hyperparamters\n",
      "{'alpha': 0.0007872452638641191, 'name': 'mlr'}\n",
      "best loss search: 0.19172079586220403\n",
      "loss refitting : 0.19172079586220403\n",
      "mlr_decade1972_lead15 already exists\n",
      "mlr_decade1982_lead15 already exists\n",
      "mlr_decade1992_lead15 already exists\n",
      "mlr_decade2002_lead15 already exists\n",
      "mlr_decade2012_lead15 already exists\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/paul/miniconda2/envs/ninolearn/lib/python3.6/site-packages/sklearn/linear_model/coordinate_descent.py:475: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations. Duality gap: 159.33193203632695, tolerance: 0.04285599581151833\n",
      "  positive)\n",
      "/home/paul/miniconda2/envs/ninolearn/lib/python3.6/site-packages/sklearn/linear_model/coordinate_descent.py:475: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations. Duality gap: 169.4946175965277, tolerance: 0.04285599581151833\n",
      "  positive)\n",
      "/home/paul/miniconda2/envs/ninolearn/lib/python3.6/site-packages/sklearn/linear_model/coordinate_descent.py:475: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations. Duality gap: 157.01499701626796, tolerance: 0.04285599581151833\n",
      "  positive)\n",
      "/home/paul/miniconda2/envs/ninolearn/lib/python3.6/site-packages/sklearn/linear_model/coordinate_descent.py:475: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations. Duality gap: 173.04389211223318, tolerance: 0.042850173842105266\n",
      "  positive)\n"
     ]
    }
   ],
   "source": [
    "from ninolearn.learn.fit import cross_training\n",
    "cross_training(mlr, pipeline, 10, alpha=[0.,0.001],  name='mlr')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Make hindcast\n",
    "\n",
    "Now, each model makes the forecast for the decade on which it was NOT trained by the function cross_training()."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "##################################################################\n",
      "Lead time: 0 months\n",
      "##################################################################\n",
      "\n",
      "Predict: 1963-01-01 till 1971-12-01\n",
      "--------------------------------------\n",
      "Predict: 1972-01-01 till 1981-12-01\n",
      "--------------------------------------\n",
      "Predict: 1982-01-01 till 1991-12-01\n",
      "--------------------------------------\n",
      "Predict: 1992-01-01 till 2001-12-01\n",
      "--------------------------------------\n",
      "Predict: 2002-01-01 till 2011-12-01\n",
      "--------------------------------------\n",
      "Predict: 2012-01-01 till 2017-12-01\n",
      "--------------------------------------\n",
      "\n",
      "##################################################################\n",
      "Lead time: 3 months\n",
      "##################################################################\n",
      "\n",
      "Predict: 1963-01-01 till 1971-12-01\n",
      "--------------------------------------\n",
      "Predict: 1972-01-01 till 1981-12-01\n",
      "--------------------------------------\n",
      "Predict: 1982-01-01 till 1991-12-01\n",
      "--------------------------------------\n",
      "Predict: 1992-01-01 till 2001-12-01\n",
      "--------------------------------------\n",
      "Predict: 2002-01-01 till 2011-12-01\n",
      "--------------------------------------\n",
      "Predict: 2012-01-01 till 2017-12-01\n",
      "--------------------------------------\n",
      "\n",
      "##################################################################\n",
      "Lead time: 6 months\n",
      "##################################################################\n",
      "\n",
      "Predict: 1963-01-01 till 1971-12-01\n",
      "--------------------------------------\n",
      "Predict: 1972-01-01 till 1981-12-01\n",
      "--------------------------------------\n",
      "Predict: 1982-01-01 till 1991-12-01\n",
      "--------------------------------------\n",
      "Predict: 1992-01-01 till 2001-12-01\n",
      "--------------------------------------\n",
      "Predict: 2002-01-01 till 2011-12-01\n",
      "--------------------------------------\n",
      "Predict: 2012-01-01 till 2017-12-01\n",
      "--------------------------------------\n",
      "\n",
      "##################################################################\n",
      "Lead time: 9 months\n",
      "##################################################################\n",
      "\n",
      "Predict: 1963-01-01 till 1971-12-01\n",
      "--------------------------------------\n",
      "Predict: 1972-01-01 till 1981-12-01\n",
      "--------------------------------------\n",
      "Predict: 1982-01-01 till 1991-12-01\n",
      "--------------------------------------\n",
      "Predict: 1992-01-01 till 2001-12-01\n",
      "--------------------------------------\n",
      "Predict: 2002-01-01 till 2011-12-01\n",
      "--------------------------------------\n",
      "Predict: 2012-01-01 till 2017-12-01\n",
      "--------------------------------------\n",
      "\n",
      "##################################################################\n",
      "Lead time: 12 months\n",
      "##################################################################\n",
      "\n",
      "Predict: 1963-01-01 till 1971-12-01\n",
      "--------------------------------------\n",
      "Predict: 1972-01-01 till 1981-12-01\n",
      "--------------------------------------\n",
      "Predict: 1982-01-01 till 1991-12-01\n",
      "--------------------------------------\n",
      "Predict: 1992-01-01 till 2001-12-01\n",
      "--------------------------------------\n",
      "Predict: 2002-01-01 till 2011-12-01\n",
      "--------------------------------------\n",
      "Predict: 2012-01-01 till 2017-12-01\n",
      "--------------------------------------\n",
      "\n",
      "##################################################################\n",
      "Lead time: 15 months\n",
      "##################################################################\n",
      "\n",
      "Predict: 1963-01-01 till 1971-12-01\n",
      "--------------------------------------\n",
      "Predict: 1972-01-01 till 1981-12-01\n",
      "--------------------------------------\n",
      "Predict: 1982-01-01 till 1991-12-01\n",
      "--------------------------------------\n",
      "Predict: 1992-01-01 till 2001-12-01\n",
      "--------------------------------------\n",
      "Predict: 2002-01-01 till 2011-12-01\n",
      "--------------------------------------\n",
      "Predict: 2012-01-01 till 2017-12-01\n",
      "--------------------------------------\n"
     ]
    }
   ],
   "source": [
    "from ninolearn.learn.fit import cross_hindcast\n",
    "cross_hindcast(mlr, pipeline, 'mlr')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluation\n",
    "\n",
    "Finally the model can be evaluated using the Pearson correlation and she standardized root-mean-squarred error (SRMSE). The SRMSE is the RMSE that is divided by the standard deviation of each season. This skill measure needs to be used instead of the RMSE  because the ONI has a seasonal cycle of the standard deviation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib.ticker import MaxNLocator\n",
    "\n",
    "\n",
    "from ninolearn.learn.fit import n_decades, lead_times, decade_color, decade_name\n",
    "from ninolearn.learn.evaluation import evaluation_correlation, evaluation_decadal_correlation, evaluation_seasonal_correlation\n",
    "from ninolearn.learn.evaluation import evaluation_srmse, evaluation_decadal_srmse, evaluation_seasonal_srmse\n",
    "from ninolearn.plot.evaluation import plot_seasonal_skill"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 468x252 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# =============================================================================\n",
    "# All season correlation skill\n",
    "# =============================================================================\n",
    "\n",
    "plt.close(\"all\")\n",
    "# scores on the full time series\n",
    "r, p  = evaluation_correlation('mlr', variable_name='prediction')\n",
    "\n",
    "# score in different decades\n",
    "r_dec, p_dec = evaluation_decadal_correlation('mlr', variable_name='prediction')\n",
    "\n",
    "# plot correlation skills\n",
    "ax = plt.figure(figsize=(6.5,3.5)).gca()\n",
    "\n",
    "for j in range(n_decades-1):\n",
    "    plt.plot(lead_times, r_dec[:,j], c=decade_color[j], label=f\"Deep Ens.  ({decade_name[j]})\")\n",
    "plt.plot(lead_times, r, label=\"Deep Ens.  (1962-2017)\", c='k', lw=2)\n",
    "\n",
    "plt.ylim(0,1)\n",
    "plt.xlim(0.,lead_times[-1])\n",
    "plt.xlabel('Lead Time [Months]')\n",
    "plt.ylabel('r')\n",
    "plt.grid()\n",
    "plt.legend(loc='center left', bbox_to_anchor=(1, 0.5))\n",
    "ax.xaxis.set_major_locator(MaxNLocator(integer=True))\n",
    "plt.tight_layout()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 468x252 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# =============================================================================\n",
    "# All season SRMSE skill\n",
    "# =============================================================================\n",
    "srmse_dec = evaluation_decadal_srmse('mlr', variable_name='prediction')\n",
    "srmse = evaluation_srmse('mlr', variable_name='prediction')\n",
    "\n",
    "# plot SRMSE skills\n",
    "ax = plt.figure(figsize=(6.5,3.5)).gca()\n",
    "for j in range(n_decades-1):\n",
    "    plt.plot(lead_times, srmse_dec[:,j], c=decade_color[j], label=f\"Deep Ens.  ({decade_name[j]})\")\n",
    "plt.plot(lead_times, srmse, label=\"Deep Ens.  (1962-2017)\", c='k', lw=2)\n",
    "\n",
    "plt.ylim(0,1.5)\n",
    "plt.xlim(0.,lead_times[-1])\n",
    "plt.xlabel('Lead Time [Months]')\n",
    "plt.ylabel('SRMSE')\n",
    "plt.grid()\n",
    "plt.legend(loc='center left', bbox_to_anchor=(1, 0.5))\n",
    "ax.xaxis.set_major_locator(MaxNLocator(integer=True))\n",
    "plt.tight_layout()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x252 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x252 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# =============================================================================\n",
    "# Seasonal skills\n",
    "# =============================================================================\n",
    "# evaluate the model in different seasons\n",
    "r_seas, p_seas = evaluation_seasonal_correlation('mlr', variable_name='prediction')\n",
    "\n",
    "plot_seasonal_skill(lead_times, r_seas,  vmin=0, vmax=1)\n",
    "plt.contour(np.arange(1,13),lead_times, p_seas, levels=[0.01, 0.05, 0.1], linestyles=['solid', 'dashed', 'dotted'], colors='k')\n",
    "plt.title('Correlation skill')\n",
    "plt.tight_layout()\n",
    "\n",
    "srsme_seas = evaluation_seasonal_srmse('mlr', variable_name='prediction')\n",
    "plot_seasonal_skill(lead_times, srsme_seas,  vmin=0, vmax=1.5, cmap=plt.cm.inferno_r, extend='max')\n",
    "plt.title('SRMSE skill')\n",
    "plt.tight_layout()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}