{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "# Train ML model to correct predictions of week 3-4 & 5-6\n", "\n", "This notebook create a Machine Learning `ML_model` to predict weeks 3-4 & 5-6 based on `S2S` weeks 3-4 & 5-6 forecasts and is compared to `CPC` observations for the [`s2s-ai-challenge`](https://s2s-ai-challenge.github.io/)." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Synopsis" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Method: `ML-based mean bias reduction`\n", "\n", "- calculate the ML-based bias from 2000-2019 deterministic ensemble mean forecast\n", "- remove that the ML-based bias from 2020 forecast deterministic ensemble mean forecast" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Data used\n", "\n", "type: renku datasets\n", "\n", "Training-input for Machine Learning model:\n", "- hindcasts of models:\n", " - ECMWF: `ecmwf_hindcast-input_2000-2019_biweekly_deterministic.zarr`\n", "\n", "Forecast-input for Machine Learning model:\n", "- real-time 2020 forecasts of models:\n", " - ECMWF: `ecmwf_forecast-input_2020_biweekly_deterministic.zarr`\n", "\n", "Compare Machine Learning model forecast against against ground truth:\n", "- `CPC` observations:\n", " - `hindcast-like-observations_biweekly_deterministic.zarr`\n", " - `forecast-like-observations_2020_biweekly_deterministic.zarr`" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Resources used\n", "for training, details in reproducibility\n", "\n", "- platform: renku\n", "- memory: 8 GB\n", "- processors: 2 CPU\n", "- storage required: 10 GB" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Safeguards\n", "\n", "All points have to be [x] checked. If not, your submission is invalid.\n", "\n", "Changes to the code after submissions are not possible, as the `commit` before the `tag` will be reviewed.\n", "(Only in exceptions and if previous effort in reproducibility can be found, it may be allowed to improve readability and reproducibility after November 1st 2021.)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Safeguards to prevent [overfitting](https://en.wikipedia.org/wiki/Overfitting?wprov=sfti1) \n", "\n", "If the organizers suspect overfitting, your contribution can be disqualified.\n", "\n", " - [x] We did not use 2020 observations in training (explicit overfitting and cheating)\n", " - [x] We did not repeatedly verify my model on 2020 observations and incrementally improved my RPSS (implicit overfitting)\n", " - [x] We provide RPSS scores for the training period with script `print_RPS_per_year`, see in section 6.3 `predict`.\n", " - [x] We tried our best to prevent [data leakage](https://en.wikipedia.org/wiki/Leakage_(machine_learning)?wprov=sfti1).\n", " - [x] We honor the `train-validate-test` [split principle](https://en.wikipedia.org/wiki/Training,_validation,_and_test_sets). This means that the hindcast data is split into `train` and `validate`, whereas `test` is withheld.\n", " - [x] We did not use `test` explicitly in training or implicitly in incrementally adjusting parameters.\n", " - [x] We considered [cross-validation](https://en.wikipedia.org/wiki/Cross-validation_(statistics))." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Safeguards for Reproducibility\n", "Notebook/code must be independently reproducible from scratch by the organizers (after the competition), if not possible: no prize\n", " - [x] All training data is publicly available (no pre-trained private neural networks, as they are not reproducible for us)\n", " - [x] Code is well documented, readable and reproducible.\n", " - [x] Code to reproduce training and predictions is preferred to run within a day on the described architecture. If the training takes longer than a day, please justify why this is needed. Please do not submit training piplelines, which take weeks to train." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Todos to improve template\n", "\n", "This is just a demo.\n", "\n", "- [ ] use multiple predictor variables and two predicted variables\n", "- [ ] for both `lead_time`s in one go\n", "- [ ] consider seasonality, for now all `forecast_time` months are mixed\n", "- [ ] make probabilistic predictions with `category` dim, for now works deterministic" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Imports" ] }, { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "from tensorflow.keras.layers import Input, Dense, Flatten\n", "from tensorflow.keras.models import Sequential\n", "\n", "import matplotlib.pyplot as plt\n", "\n", "import xarray as xr\n", "xr.set_options(display_style='text')\n", "import numpy as np\n", "\n", "from dask.utils import format_bytes\n", "import xskillscore as xs" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Get training data\n", "\n", "preprocessing of input data may be done in separate notebook/script" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Hindcast\n", "\n", "get weekly initialized hindcasts" ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [], "source": [ "v='t2m'" ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\u001b[33m\u001b[1mWarning: \u001b[0mRun CLI commands only from project's root directory.\n", "\u001b[0m\n" ] } ], "source": [ "# preprocessed as renku dataset\n", "!renku storage pull ../data/ecmwf_hindcast-input_2000-2019_biweekly_deterministic.zarr" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [], "source": [ "hind_2000_2019 = xr.open_zarr(\"../data/ecmwf_hindcast-input_2000-2019_biweekly_deterministic.zarr\", consolidated=True)" ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\u001b[33m\u001b[1mWarning: \u001b[0mRun CLI commands only from project's root directory.\n", "\u001b[0m\n" ] } ], "source": [ "# preprocessed as renku dataset\n", "!renku storage pull ../data/ecmwf_forecast-input_2020_biweekly_deterministic.zarr" ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [], "source": [ "fct_2020 = xr.open_zarr(\"../data/ecmwf_forecast-input_2020_biweekly_deterministic.zarr\", consolidated=True)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Observations\n", "corresponding to hindcasts" ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\u001b[33m\u001b[1mWarning: \u001b[0mRun CLI commands only from project's root directory.\n", "\u001b[0m\n" ] } ], "source": [ "# preprocessed as renku dataset\n", "!renku storage pull ../data/hindcast-like-observations_2000-2019_biweekly_deterministic.zarr" ] }, { "cell_type": "code", "execution_count": 8, "metadata": {}, "outputs": [], "source": [ "obs_2000_2019 = xr.open_zarr(\"../data/hindcast-like-observations_2000-2019_biweekly_deterministic.zarr\", consolidated=True)#[v]" ] }, { "cell_type": "code", "execution_count": 9, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\u001b[33m\u001b[1mWarning: \u001b[0mRun CLI commands only from project's root directory.\n", "\u001b[0m\n" ] } ], "source": [ "# preprocessed as renku dataset\n", "!renku storage pull ../data/forecast-like-observations_2020_biweekly_deterministic.zarr" ] }, { "cell_type": "code", "execution_count": 10, "metadata": {}, "outputs": [], "source": [ "obs_2020 = xr.open_zarr(\"../data/forecast-like-observations_2020_biweekly_deterministic.zarr\", consolidated=True)#[v]" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# ML model" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "based on [Weatherbench](https://github.com/pangeo-data/WeatherBench/blob/master/quickstart.ipynb)" ] }, { "cell_type": "code", "execution_count": 11, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "fatal: destination path 'WeatherBench' already exists and is not an empty directory.\n" ] } ], "source": [ "# run once only and dont commit\n", "!git clone https://github.com/pangeo-data/WeatherBench/" ] }, { "cell_type": "code", "execution_count": 12, "metadata": {}, "outputs": [], "source": [ "import sys\n", "sys.path.insert(1, 'WeatherBench')\n", "from WeatherBench.src.train_nn import DataGenerator, PeriodicConv2D, create_predictions\n", "import tensorflow.keras as keras" ] }, { "cell_type": "code", "execution_count": 13, "metadata": {}, "outputs": [], "source": [ "bs=32\n", "\n", "import numpy as np\n", "class DataGenerator(keras.utils.Sequence):\n", " def __init__(self, fct, verif, lead_time, batch_size=bs, shuffle=True, load=True,\n", " mean=None, std=None):\n", " \"\"\"\n", " Data generator for WeatherBench data.\n", " Template from https://stanford.edu/~shervine/blog/keras-how-to-generate-data-on-the-fly\n", "\n", " Args:\n", " fct: forecasts from S2S models: xr.DataArray (xr.Dataset doesnt work properly)\n", " verif: observations with same dimensionality (xr.Dataset doesnt work properly)\n", " lead_time: Lead_time as in model\n", " batch_size: Batch size\n", " shuffle: bool. If True, data is shuffled.\n", " load: bool. If True, datadet is loaded into RAM.\n", " mean: If None, compute mean from data.\n", " std: If None, compute standard deviation from data.\n", " \n", " Todo:\n", " - use number in a better way, now uses only ensemble mean forecast\n", " - dont use .sel(lead_time=lead_time) to train over all lead_time at once\n", " - be sensitive with forecast_time, pool a few around the weekofyear given\n", " - use more variables as predictors\n", " - predict more variables\n", " \"\"\"\n", "\n", " if isinstance(fct, xr.Dataset):\n", " print('convert fct to array')\n", " fct = fct.to_array().transpose(...,'variable')\n", " self.fct_dataset=True\n", " else:\n", " self.fct_dataset=False\n", " \n", " if isinstance(verif, xr.Dataset):\n", " print('convert verif to array')\n", " verif = verif.to_array().transpose(...,'variable')\n", " self.verif_dataset=True\n", " else:\n", " self.verif_dataset=False\n", " \n", " #self.fct = fct\n", " self.batch_size = batch_size\n", " self.shuffle = shuffle\n", " self.lead_time = lead_time\n", "\n", " self.fct_data = fct.transpose('forecast_time', ...).sel(lead_time=lead_time)\n", " self.fct_mean = self.fct_data.mean('forecast_time').compute() if mean is None else mean\n", " self.fct_std = self.fct_data.std('forecast_time').compute() if std is None else std\n", " \n", " self.verif_data = verif.transpose('forecast_time', ...).sel(lead_time=lead_time)\n", " self.verif_mean = self.verif_data.mean('forecast_time').compute() if mean is None else mean\n", " self.verif_std = self.verif_data.std('forecast_time').compute() if std is None else std\n", "\n", " # Normalize\n", " self.fct_data = (self.fct_data - self.fct_mean) / self.fct_std\n", " self.verif_data = (self.verif_data - self.verif_mean) / self.verif_std\n", " \n", " self.n_samples = self.fct_data.forecast_time.size\n", " self.forecast_time = self.fct_data.forecast_time\n", "\n", " self.on_epoch_end()\n", "\n", " # For some weird reason calling .load() earlier messes up the mean and std computations\n", " if load:\n", " # print('Loading data into RAM')\n", " self.fct_data.load()\n", "\n", " def __len__(self):\n", " 'Denotes the number of batches per epoch'\n", " return int(np.ceil(self.n_samples / self.batch_size))\n", "\n", " def __getitem__(self, i):\n", " 'Generate one batch of data'\n", " idxs = self.idxs[i * self.batch_size:(i + 1) * self.batch_size]\n", " # got all nan if nans not masked\n", " X = self.fct_data.isel(forecast_time=idxs).fillna(0.).values\n", " y = self.verif_data.isel(forecast_time=idxs).fillna(0.).values\n", " return X, y\n", "\n", " def on_epoch_end(self):\n", " 'Updates indexes after each epoch'\n", " self.idxs = np.arange(self.n_samples)\n", " if self.shuffle == True:\n", " np.random.shuffle(self.idxs)" ] }, { "cell_type": "code", "execution_count": 14, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
<xarray.DataArray 'lead_time' ()>\n",
       "array(1209600000000000, dtype='timedelta64[ns]')\n",
       "Coordinates:\n",
       "    lead_time  timedelta64[ns] 14 days\n",
       "Attributes:\n",
       "    aggregate:      The pd.Timedelta corresponds to the first day of a biweek...\n",
       "    description:    Forecast period is the time interval between the forecast...\n",
       "    long_name:      lead time\n",
       "    standard_name:  forecast_period\n",
       "    week34_t2m:     mean[14 days, 27 days]\n",
       "    week34_tp:      28 days minus 14 days\n",
       "    week56_t2m:     mean[28 days, 41 days]\n",
       "    week56_tp:      42 days minus 28 days
" ], "text/plain": [ "\n", "array(1209600000000000, dtype='timedelta64[ns]')\n", "Coordinates:\n", " lead_time timedelta64[ns] 14 days\n", "Attributes:\n", " aggregate: The pd.Timedelta corresponds to the first day of a biweek...\n", " description: Forecast period is the time interval between the forecast...\n", " long_name: lead time\n", " standard_name: forecast_period\n", " week34_t2m: mean[14 days, 27 days]\n", " week34_tp: 28 days minus 14 days\n", " week56_t2m: mean[28 days, 41 days]\n", " week56_tp: 42 days minus 28 days" ] }, "execution_count": 14, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# 2 bi-weekly `lead_time`: week 3-4\n", "lead = hind_2000_2019.isel(lead_time=0).lead_time\n", "\n", "lead" ] }, { "cell_type": "code", "execution_count": 15, "metadata": {}, "outputs": [], "source": [ "# mask, needed?\n", "hind_2000_2019 = hind_2000_2019.where(obs_2000_2019.isel(forecast_time=0, lead_time=0,drop=True).notnull())" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## data prep: train, valid, test\n", "\n", "[Use the hindcast period to split train and valid.](https://en.wikipedia.org/wiki/Training,_validation,_and_test_sets) Do not use the 2020 data for testing!" ] }, { "cell_type": "code", "execution_count": 16, "metadata": {}, "outputs": [], "source": [ "# time is the forecast_time\n", "time_train_start,time_train_end='2000','2017' # train\n", "time_valid_start,time_valid_end='2018','2019' # valid\n", "time_test = '2020' # test" ] }, { "cell_type": "code", "execution_count": 17, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "/opt/conda/lib/python3.8/site-packages/dask/array/numpy_compat.py:39: RuntimeWarning: invalid value encountered in true_divide\n", " x = np.divide(x1, x2, out)\n", "/opt/conda/lib/python3.8/site-packages/dask/array/numpy_compat.py:39: RuntimeWarning: invalid value encountered in true_divide\n", " x = np.divide(x1, x2, out)\n", "/opt/conda/lib/python3.8/site-packages/dask/array/numpy_compat.py:39: RuntimeWarning: invalid value encountered in true_divide\n", " x = np.divide(x1, x2, out)\n", "/opt/conda/lib/python3.8/site-packages/dask/array/numpy_compat.py:39: RuntimeWarning: invalid value encountered in true_divide\n", " x = np.divide(x1, x2, out)\n", "/opt/conda/lib/python3.8/site-packages/dask/array/numpy_compat.py:39: RuntimeWarning: invalid value encountered in true_divide\n", " x = np.divide(x1, x2, out)\n", "/opt/conda/lib/python3.8/site-packages/dask/array/numpy_compat.py:39: RuntimeWarning: invalid value encountered in true_divide\n", " x = np.divide(x1, x2, out)\n" ] } ], "source": [ "dg_train = DataGenerator(\n", " hind_2000_2019.mean('realization').sel(forecast_time=slice(time_train_start,time_train_end))[v],\n", " obs_2000_2019.sel(forecast_time=slice(time_train_start,time_train_end))[v],\n", " lead_time=lead, batch_size=bs, load=True)" ] }, { "cell_type": "code", "execution_count": 18, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "/opt/conda/lib/python3.8/site-packages/dask/array/numpy_compat.py:39: RuntimeWarning: invalid value encountered in true_divide\n", " x = np.divide(x1, x2, out)\n", "/opt/conda/lib/python3.8/site-packages/dask/array/numpy_compat.py:39: RuntimeWarning: invalid value encountered in true_divide\n", " x = np.divide(x1, x2, out)\n", "/opt/conda/lib/python3.8/site-packages/dask/array/numpy_compat.py:39: RuntimeWarning: invalid value encountered in true_divide\n", " x = np.divide(x1, x2, out)\n", "/opt/conda/lib/python3.8/site-packages/dask/array/numpy_compat.py:39: RuntimeWarning: invalid value encountered in true_divide\n", " x = np.divide(x1, x2, out)\n", "/opt/conda/lib/python3.8/site-packages/dask/array/numpy_compat.py:39: RuntimeWarning: invalid value encountered in true_divide\n", " x = np.divide(x1, x2, out)\n" ] } ], "source": [ "dg_valid = DataGenerator(\n", " hind_2000_2019.mean('realization').sel(forecast_time=slice(time_valid_start,time_valid_end))[v],\n", " obs_2000_2019.sel(forecast_time=slice(time_valid_start,time_valid_end))[v],\n", " lead_time=lead, batch_size=bs, shuffle=False, load=True)" ] }, { "cell_type": "code", "execution_count": 19, "metadata": {}, "outputs": [], "source": [ "# do not use, delete?\n", "dg_test = DataGenerator(\n", " fct_2020.mean('realization').sel(forecast_time=time_test)[v],\n", " obs_2020.sel(forecast_time=time_test)[v],\n", " lead_time=lead, batch_size=bs, load=True, mean=dg_train.fct_mean, std=dg_train.fct_std, shuffle=False)" ] }, { "cell_type": "code", "execution_count": 20, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "((32, 121, 240), (32, 121, 240))" ] }, "execution_count": 20, "metadata": {}, "output_type": "execute_result" } ], "source": [ "X, y = dg_valid[0]\n", "X.shape, y.shape" ] }, { "cell_type": "code", "execution_count": 21, "metadata": {}, "outputs": [], "source": [ "# short look into training data: large biases\n", "# any problem from normalizing?\n", "# i=4\n", "# xr.DataArray(np.vstack([X[i],y[i]])).plot(yincrease=False, robust=True)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## `fit`" ] }, { "cell_type": "code", "execution_count": 22, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "WARNING:tensorflow:AutoGraph could not transform > and will run it as-is.\n", "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n", "Cause: module 'gast' has no attribute 'Index'\n", "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n", "WARNING: AutoGraph could not transform > and will run it as-is.\n", "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n", "Cause: module 'gast' has no attribute 'Index'\n", "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n" ] } ], "source": [ "cnn = keras.models.Sequential([\n", " PeriodicConv2D(filters=32, kernel_size=5, conv_kwargs={'activation':'relu'}, input_shape=(32, 64, 1)),\n", " PeriodicConv2D(filters=1, kernel_size=5)\n", "])" ] }, { "cell_type": "code", "execution_count": 23, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Model: \"sequential\"\n", "_________________________________________________________________\n", "Layer (type) Output Shape Param # \n", "=================================================================\n", "periodic_conv2d (PeriodicCon (None, 32, 64, 32) 832 \n", "_________________________________________________________________\n", "periodic_conv2d_1 (PeriodicC (None, 32, 64, 1) 801 \n", "=================================================================\n", "Total params: 1,633\n", "Trainable params: 1,633\n", "Non-trainable params: 0\n", "_________________________________________________________________\n" ] } ], "source": [ "cnn.summary()" ] }, { "cell_type": "code", "execution_count": 24, "metadata": {}, "outputs": [], "source": [ "cnn.compile(keras.optimizers.Adam(1e-4), 'mse')" ] }, { "cell_type": "code", "execution_count": 25, "metadata": {}, "outputs": [], "source": [ "import warnings\n", "warnings.simplefilter(\"ignore\")" ] }, { "cell_type": "code", "execution_count": 26, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Epoch 1/2\n", "30/30 [==============================] - 58s 2s/step - loss: 0.1472 - val_loss: 0.0742\n", "Epoch 2/2\n", "30/30 [==============================] - 45s 1s/step - loss: 0.0712 - val_loss: 0.0545\n" ] }, { "data": { "text/plain": [ "" ] }, "execution_count": 26, "metadata": {}, "output_type": "execute_result" } ], "source": [ "cnn.fit(dg_train, epochs=2, validation_data=dg_valid)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## `predict`\n", "\n", "Create predictions and print `mean(variable, lead_time, longitude, weighted latitude)` RPSS for all years as calculated by `skill_by_year`." ] }, { "cell_type": "code", "execution_count": 27, "metadata": {}, "outputs": [], "source": [ "from scripts import add_valid_time_from_forecast_reference_time_and_lead_time\n", "\n", "def _create_predictions(model, dg, lead):\n", " \"\"\"Create non-iterative predictions\"\"\"\n", " preds = model.predict(dg).squeeze()\n", " # Unnormalize\n", " preds = preds * dg.fct_std.values + dg.fct_mean.values\n", " if dg.verif_dataset:\n", " da = xr.DataArray(\n", " preds,\n", " dims=['forecast_time', 'latitude', 'longitude','variable'],\n", " coords={'forecast_time': dg.fct_data.forecast_time, 'latitude': dg.fct_data.latitude,\n", " 'longitude': dg.fct_data.longitude},\n", " ).to_dataset() # doesnt work yet\n", " else:\n", " da = xr.DataArray(\n", " preds,\n", " dims=['forecast_time', 'latitude', 'longitude'],\n", " coords={'forecast_time': dg.fct_data.forecast_time, 'latitude': dg.fct_data.latitude,\n", " 'longitude': dg.fct_data.longitude},\n", " )\n", " da = da.assign_coords(lead_time=lead)\n", " # da = add_valid_time_from_forecast_reference_time_and_lead_time(da)\n", " return da" ] }, { "cell_type": "code", "execution_count": 28, "metadata": {}, "outputs": [], "source": [ "# optionally masking the ocean when making probabilistic\n", "mask = obs_2020.std(['lead_time','forecast_time']).notnull()" ] }, { "cell_type": "code", "execution_count": 29, "metadata": {}, "outputs": [], "source": [ "from scripts import make_probabilistic" ] }, { "cell_type": "code", "execution_count": 30, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\u001b[33m\u001b[1mWarning: \u001b[0mRun CLI commands only from project's root directory.\n", "\u001b[0m\n" ] } ], "source": [ "!renku storage pull ../data/hindcast-like-observations_2000-2019_biweekly_tercile-edges.nc" ] }, { "cell_type": "code", "execution_count": 31, "metadata": {}, "outputs": [], "source": [ "cache_path='../data'\n", "tercile_file = f'{cache_path}/hindcast-like-observations_2000-2019_biweekly_tercile-edges.nc'\n", "tercile_edges = xr.open_dataset(tercile_file)" ] }, { "cell_type": "code", "execution_count": 32, "metadata": {}, "outputs": [], "source": [ "# this is not useful but results have expected dimensions\n", "# actually train for each lead_time\n", "\n", "def create_predictions(cnn, fct, obs, time):\n", " preds_test=[]\n", " for lead in fct.lead_time:\n", " dg = DataGenerator(fct.mean('realization').sel(forecast_time=time)[v],\n", " obs.sel(forecast_time=time)[v],\n", " lead_time=lead, batch_size=bs, mean=dg_train.fct_mean, std=dg_train.fct_std, shuffle=False)\n", " preds_test.append(_create_predictions(cnn, dg, lead))\n", " preds_test = xr.concat(preds_test, 'lead_time')\n", " preds_test['lead_time'] = fct.lead_time\n", " # add valid_time coord\n", " preds_test = add_valid_time_from_forecast_reference_time_and_lead_time(preds_test)\n", " preds_test = preds_test.to_dataset(name=v)\n", " # add fake var\n", " preds_test['tp'] = preds_test['t2m']\n", " # make probabilistic\n", " preds_test = make_probabilistic(preds_test.expand_dims('realization'), tercile_edges, mask=mask)\n", " return preds_test" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### `predict` training period in-sample" ] }, { "cell_type": "code", "execution_count": 33, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\u001b[33m\u001b[1mWarning: \u001b[0mRun CLI commands only from project's root directory.\n", "\u001b[0m\n" ] } ], "source": [ "!renku storage pull ../data/forecast-like-observations_2020_biweekly_terciled.nc" ] }, { "cell_type": "code", "execution_count": 34, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\u001b[33m\u001b[1mWarning: \u001b[0mRun CLI commands only from project's root directory.\n", "\u001b[0m\n" ] } ], "source": [ "!renku storage pull ../data/hindcast-like-observations_2000-2019_biweekly_terciled.zarr" ] }, { "cell_type": "code", "execution_count": 35, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
RPSS
year
2000-0.862483
2001-1.015485
2002-1.101022
2003-1.032647
2004-1.056348
2005-1.165675
2006-1.057217
2007-1.170849
2008-1.049785
2009-1.169108
2010-1.130845
2011-1.052670
2012-1.126449
2013-1.126930
2014-1.095896
2015-1.117486
\n", "
" ], "text/plain": [ " RPSS\n", "year \n", "2000 -0.862483\n", "2001 -1.015485\n", "2002 -1.101022\n", "2003 -1.032647\n", "2004 -1.056348\n", "2005 -1.165675\n", "2006 -1.057217\n", "2007 -1.170849\n", "2008 -1.049785\n", "2009 -1.169108\n", "2010 -1.130845\n", "2011 -1.052670\n", "2012 -1.126449\n", "2013 -1.126930\n", "2014 -1.095896\n", "2015 -1.117486" ] }, "execution_count": 35, "metadata": {}, "output_type": "execute_result" } ], "source": [ "from scripts import skill_by_year\n", "import os\n", "if os.environ['HOME'] == '/home/jovyan':\n", " import pandas as pd\n", " # assume on renku with small memory\n", " step = 2\n", " skill_list = []\n", " for year in np.arange(int(time_train_start), int(time_train_end) -1, step): # loop over years to consume less memory on renku\n", " preds_is = create_predictions(cnn, hind_2000_2019, obs_2000_2019, time=slice(str(year), str(year+step-1))).compute()\n", " skill_list.append(skill_by_year(preds_is))\n", " skill = pd.concat(skill_list)\n", "else: # with larger memory, simply do\n", " preds_is = create_predictions(cnn, hind_2000_2019, obs_2000_2019, time=slice(time_train_start, time_train_end))\n", " skill = skill_by_year(preds_is)\n", "skill" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### `predict` validation period out-of-sample" ] }, { "cell_type": "code", "execution_count": 36, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
RPSS
year
2018-1.099744
2019-1.172401
\n", "
" ], "text/plain": [ " RPSS\n", "year \n", "2018 -1.099744\n", "2019 -1.172401" ] }, "execution_count": 36, "metadata": {}, "output_type": "execute_result" } ], "source": [ "preds_os = create_predictions(cnn, hind_2000_2019, obs_2000_2019, time=slice(time_valid_start, time_valid_end))\n", "\n", "skill_by_year(preds_os)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### `predict` test" ] }, { "cell_type": "code", "execution_count": 37, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
RPSS
year
2020-1.076834
\n", "
" ], "text/plain": [ " RPSS\n", "year \n", "2020 -1.076834" ] }, "execution_count": 37, "metadata": {}, "output_type": "execute_result" } ], "source": [ "preds_test = create_predictions(cnn, fct_2020, obs_2020, time=time_test)\n", "\n", "skill_by_year(preds_test)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Submission" ] }, { "cell_type": "code", "execution_count": 38, "metadata": {}, "outputs": [], "source": [ "from scripts import assert_predictions_2020\n", "assert_predictions_2020(preds_test)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "preds_test.to_netcdf('../submissions/ML_prediction_2020.nc')" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "# !git add ../submissions/ML_prediction_2020.nc\n", "# !git add ML_train_and_prediction.ipynb" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "# !git commit -m \"template_test commit message\" # whatever message you want" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "# !git tag \"submission-template_test-0.0.1\" # if this is to be checked by scorer, only the last submitted==tagged version will be considered" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "# !git push --tags" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Reproducibility" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## memory" ] }, { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ " total used free shared buff/cache available\n", "Mem: 31 7 11 0 12 24\n", "Swap: 0 0 0\n" ] } ], "source": [ "# https://phoenixnap.com/kb/linux-commands-check-memory-usage\n", "!free -g" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## CPU" ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Architecture: x86_64\n", "CPU op-mode(s): 32-bit, 64-bit\n", "Byte Order: Little Endian\n", "Address sizes: 40 bits physical, 48 bits virtual\n", "CPU(s): 8\n", "On-line CPU(s) list: 0-7\n", "Thread(s) per core: 1\n", "Core(s) per socket: 1\n", "Socket(s): 8\n", "NUMA node(s): 1\n", "Vendor ID: GenuineIntel\n", "CPU family: 6\n", "Model: 85\n", "Model name: Intel Xeon Processor (Skylake, IBRS)\n", "Stepping: 4\n", "CPU MHz: 2095.078\n", "BogoMIPS: 4190.15\n", "Virtualization: VT-x\n", "Hypervisor vendor: KVM\n", "Virtualization type: full\n", "L1d cache: 256 KiB\n", "L1i cache: 256 KiB\n", "L2 cache: 32 MiB\n", "L3 cache: 128 MiB\n", "NUMA node0 CPU(s): 0-7\n", "Vulnerability Itlb multihit: KVM: Mitigation: Split huge pages\n", "Vulnerability L1tf: Mitigation; PTE Inversion; VMX conditional cach\n", " e flushes, SMT disabled\n", "Vulnerability Mds: Vulnerable: Clear CPU buffers attempted, no mic\n", " rocode; SMT Host state unknown\n", "Vulnerability Meltdown: Mitigation; PTI\n", "Vulnerability Spec store bypass: Vulnerable\n", "Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user\n", " pointer sanitization\n", "Vulnerability Spectre v2: Mitigation; Full generic retpoline, IBPB condit\n", " ional, IBRS_FW, STIBP disabled, RSB filling\n", "Vulnerability Srbds: Not affected\n", "Vulnerability Tsx async abort: Not affected\n", "Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtr\n", " r pge mca cmov pat pse36 clflush mmx fxsr sse s\n", " se2 syscall nx pdpe1gb rdtscp lm constant_tsc r\n", " ep_good nopl xtopology cpuid tsc_known_freq pni\n", " pclmulqdq vmx ssse3 fma cx16 pcid sse4_1 sse4_\n", " 2 x2apic movbe popcnt tsc_deadline_timer aes xs\n", " ave avx f16c rdrand hypervisor lahf_lm abm 3dno\n", " wprefetch cpuid_fault invpcid_single pti ibrs i\n", " bpb tpr_shadow vnmi flexpriority ept vpid ept_a\n", " d fsgsbase bmi1 avx2 smep bmi2 erms invpcid avx\n", " 512f avx512dq rdseed adx smap clwb avx512cd avx\n", " 512bw avx512vl xsaveopt xsavec xgetbv1 arat pku\n", " ospke\n" ] } ], "source": [ "!lscpu" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## software" ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "# packages in environment at /opt/conda:\n", "#\n", "# Name Version Build Channel\n", "_libgcc_mutex 0.1 conda_forge conda-forge\n", "_openmp_mutex 4.5 1_gnu conda-forge\n", "_pytorch_select 0.1 cpu_0 defaults\n", "_tflow_select 2.3.0 mkl defaults\n", "absl-py 0.13.0 py38h06a4308_0 defaults\n", "aiobotocore 1.4.1 pyhd3eb1b0_0 defaults\n", "aiohttp 3.7.4.post0 py38h7f8727e_2 defaults\n", "aioitertools 0.7.1 pyhd3eb1b0_0 defaults\n", "alembic 1.4.3 pyh9f0ad1d_0 conda-forge\n", "ansiwrap 0.8.4 pypi_0 pypi\n", "appdirs 1.4.4 pypi_0 pypi\n", "argcomplete 1.12.3 pypi_0 pypi\n", "argon2-cffi 20.1.0 py38h497a2fe_2 conda-forge\n", "argparse 1.4.0 pypi_0 pypi\n", "asciitree 0.3.3 py_2 defaults\n", "astor 0.8.1 py38h06a4308_0 defaults\n", "astunparse 1.6.3 py_0 defaults\n", "async-timeout 3.0.1 pypi_0 pypi\n", "async_generator 1.10 py_0 conda-forge\n", "attrs 21.2.0 pypi_0 pypi\n", "backcall 0.2.0 pyh9f0ad1d_0 conda-forge\n", "backports 1.0 py_2 conda-forge\n", "backports.functools_lru_cache 1.6.1 py_0 conda-forge\n", "bagit 1.8.1 pypi_0 pypi\n", "beautifulsoup4 4.10.0 pyh06a4308_0 defaults\n", "binutils_impl_linux-64 2.35.1 h193b22a_1 conda-forge\n", "binutils_linux-64 2.35 h67ddf6f_30 conda-forge\n", "black 20.8b1 pypi_0 pypi\n", "blas 1.0 mkl defaults\n", "bleach 3.2.1 pyh9f0ad1d_0 conda-forge\n", "blinker 1.4 py_1 conda-forge\n", "bokeh 2.3.3 py38h06a4308_0 defaults\n", "botocore 1.20.106 pyhd3eb1b0_0 defaults\n", "bottleneck 1.3.2 py38heb32a55_1 defaults\n", "bracex 2.1.1 pypi_0 pypi\n", "branca 0.3.1 pypi_0 pypi\n", "brotli 1.0.9 he6710b0_2 defaults\n", "brotlipy 0.7.0 py38h497a2fe_1001 conda-forge\n", "bzip2 1.0.8 h7f98852_4 conda-forge\n", "c-ares 1.17.1 h36c2ea0_0 conda-forge\n", "ca-certificates 2021.7.5 h06a4308_1 defaults\n", "cachecontrol 0.12.6 pypi_0 pypi\n", "cachetools 4.2.4 pypi_0 pypi\n", "calamus 0.3.12 pypi_0 pypi\n", "cdsapi 0.5.1 pypi_0 pypi\n", "certifi 2021.5.30 pypi_0 pypi\n", "certipy 0.1.3 py_0 conda-forge\n", "cffi 1.14.6 pypi_0 pypi\n", "cfgrib 0.9.9.0 pyhd8ed1ab_1 conda-forge\n", "cftime 1.5.0 py38h6323ea4_0 defaults\n", "chardet 3.0.4 pypi_0 pypi\n", "click 7.1.2 pypi_0 pypi\n", "click-completion 0.5.2 pypi_0 pypi\n", "click-option-group 0.5.3 pypi_0 pypi\n", "click-plugins 1.1.1 pypi_0 pypi\n", "climetlab 0.8.31 pypi_0 pypi\n", "climetlab-s2s-ai-challenge 0.8.0 pypi_0 pypi\n", "cloudpickle 2.0.0 pyhd3eb1b0_0 defaults\n", "colorama 0.4.4 pypi_0 pypi\n", "coloredlogs 15.0.1 pypi_0 pypi\n", "commonmark 0.9.1 pypi_0 pypi\n", "conda 4.9.2 py38h578d9bd_0 conda-forge\n", "conda-package-handling 1.7.2 py38h8df0ef7_0 conda-forge\n", "configargparse 1.5.2 pypi_0 pypi\n", "configurable-http-proxy 1.3.0 0 conda-forge\n", "coverage 5.5 py38h27cfd23_2 defaults\n", "cryptography 3.4.8 pypi_0 pypi\n", "curl 7.71.1 he644dc0_8 conda-forge\n", "cwlgen 0.4.2 pypi_0 pypi\n", "cwltool 3.1.20211004060744 pypi_0 pypi\n", "cycler 0.10.0 py38_0 defaults\n", "cython 0.29.24 py38h295c915_0 defaults\n", "cytoolz 0.11.0 py38h7b6447c_0 defaults\n", "dask 2021.8.1 pyhd3eb1b0_0 defaults\n", "dask-core 2021.8.1 pyhd3eb1b0_0 defaults\n", "dataclasses 0.8 pyh6d0b6a4_7 defaults\n", "decorator 4.4.2 py_0 conda-forge\n", "defusedxml 0.6.0 py_0 conda-forge\n", "distributed 2021.8.1 py38h06a4308_0 defaults\n", "distro 1.5.0 pypi_0 pypi\n", "docopt 0.6.2 py38h06a4308_0 defaults\n", "eccodes 2.21.0 ha0e6eb6_0 conda-forge\n", "ecmwf-api-client 1.6.1 pypi_0 pypi\n", "ecmwflibs 0.3.14 pypi_0 pypi\n", "entrypoints 0.3 pyhd8ed1ab_1003 conda-forge\n", "environ-config 21.2.0 pypi_0 pypi\n", "fasteners 0.16.3 pyhd3eb1b0_0 defaults\n", "filelock 3.0.12 pypi_0 pypi\n", "findlibs 0.0.2 pypi_0 pypi\n", "fonttools 4.25.0 pyhd3eb1b0_0 defaults\n", "freetype 2.10.4 h5ab3b9f_0 defaults\n", "frozendict 2.0.6 pypi_0 pypi\n", "fsspec 2021.7.0 pyhd3eb1b0_0 defaults\n", "gast 0.4.0 pyhd3eb1b0_0 defaults\n", "gcc_impl_linux-64 9.3.0 h70c0ae5_18 conda-forge\n", "gcc_linux-64 9.3.0 hf25ea35_30 conda-forge\n", "gitdb 4.0.7 pypi_0 pypi\n", "gitpython 3.1.14 pypi_0 pypi\n", "google-auth 1.33.0 pyhd3eb1b0_0 defaults\n", "google-auth-oauthlib 0.4.4 pyhd3eb1b0_0 defaults\n", "google-pasta 0.2.0 pyhd3eb1b0_0 defaults\n", "grpcio 1.36.1 py38h2157cd5_1 defaults\n", "gxx_impl_linux-64 9.3.0 hd87eabc_18 conda-forge\n", "gxx_linux-64 9.3.0 h3fbe746_30 conda-forge\n", "h5netcdf 0.11.0 pyhd8ed1ab_0 conda-forge\n", "h5py 2.10.0 py38hd6299e0_1 defaults\n", "hdf4 4.2.13 h3ca952b_2 defaults\n", "hdf5 1.10.6 nompi_h6a2412b_1114 conda-forge\n", "heapdict 1.0.1 pyhd3eb1b0_0 defaults\n", "humanfriendly 10.0 pypi_0 pypi\n", "humanize 3.7.1 pypi_0 pypi\n", "icu 68.1 h58526e2_0 conda-forge\n", "idna 2.10 pyh9f0ad1d_0 conda-forge\n", "importlib-metadata 3.4.0 py38h578d9bd_0 conda-forge\n", "importlib_metadata 3.4.0 hd8ed1ab_0 conda-forge\n", "intake 0.6.3 pyhd3eb1b0_0 defaults\n", "intake-xarray 0.5.0 pyhd3eb1b0_0 defaults\n", "intel-openmp 2019.4 243 defaults\n", "ipykernel 5.4.2 py38h81c977d_0 conda-forge\n", "ipython 7.19.0 py38h81c977d_2 conda-forge\n", "ipython_genutils 0.2.0 py_1 conda-forge\n", "isodate 0.6.0 pypi_0 pypi\n", "jasper 1.900.1 hd497a04_4 defaults\n", "jedi 0.17.2 py38h578d9bd_1 conda-forge\n", "jellyfish 0.8.8 pypi_0 pypi\n", "jinja2 3.0.1 pypi_0 pypi\n", "jmespath 0.10.0 pyhd3eb1b0_0 defaults\n", "joblib 1.0.1 pyhd3eb1b0_0 defaults\n", "jpeg 9d h7f8727e_0 defaults\n", "json5 0.9.5 pyh9f0ad1d_0 conda-forge\n", "jsonschema 3.2.0 py_2 conda-forge\n", "jupyter-server-proxy 1.6.0 pypi_0 pypi\n", "jupyter_client 6.1.11 pyhd8ed1ab_1 conda-forge\n", "jupyter_core 4.7.0 py38h578d9bd_0 conda-forge\n", "jupyter_telemetry 0.1.0 pyhd8ed1ab_1 conda-forge\n", "jupyterhub 1.2.2 pypi_0 pypi\n", "jupyterlab 2.2.9 py_0 conda-forge\n", "jupyterlab-git 0.23.3 pypi_0 pypi\n", "jupyterlab_pygments 0.1.2 pyh9f0ad1d_0 conda-forge\n", "jupyterlab_server 1.2.0 py_0 conda-forge\n", "keras-preprocessing 1.1.2 pyhd3eb1b0_0 defaults\n", "kernel-headers_linux-64 2.6.32 h77966d4_13 conda-forge\n", "kiwisolver 1.3.1 py38h2531618_0 defaults\n", "krb5 1.17.2 h926e7f8_0 conda-forge\n", "lazy-object-proxy 1.6.0 pypi_0 pypi\n", "lcms2 2.12 h3be6417_0 defaults\n", "ld_impl_linux-64 2.35.1 hea4e1c9_1 conda-forge\n", "libaec 1.0.4 he6710b0_1 defaults\n", "libblas 3.9.0 1_h86c2bf4_netlib conda-forge\n", "libcblas 3.9.0 5_h92ddd45_netlib conda-forge\n", "libcurl 7.71.1 hcdd3856_8 conda-forge\n", "libedit 3.1.20191231 he28a2e2_2 conda-forge\n", "libev 4.33 h516909a_1 conda-forge\n", "libffi 3.3 h58526e2_2 conda-forge\n", "libgcc-devel_linux-64 9.3.0 h7864c58_18 conda-forge\n", "libgcc-ng 9.3.0 h2828fa1_18 conda-forge\n", "libgfortran-ng 9.3.0 ha5ec8a7_17 defaults\n", "libgfortran5 9.3.0 ha5ec8a7_17 defaults\n", "libgomp 9.3.0 h2828fa1_18 conda-forge\n", "liblapack 3.9.0 5_h92ddd45_netlib conda-forge\n", "libllvm10 10.0.1 hbcb73fb_5 defaults\n", "libmklml 2019.0.5 0 defaults\n", "libnetcdf 4.7.4 nompi_h56d31a8_107 conda-forge\n", "libnghttp2 1.41.0 h8cfc5f6_2 conda-forge\n", "libpng 1.6.37 hbc83047_0 defaults\n", "libprotobuf 3.17.2 h4ff587b_1 defaults\n", "libsodium 1.0.18 h36c2ea0_1 conda-forge\n", "libssh2 1.9.0 hab1572f_5 conda-forge\n", "libstdcxx-devel_linux-64 9.3.0 hb016644_18 conda-forge\n", "libstdcxx-ng 9.3.0 h6de172a_18 conda-forge\n", "libtiff 4.2.0 h85742a9_0 defaults\n", "libuv 1.40.0 h7f98852_0 conda-forge\n", "libwebp-base 1.2.0 h27cfd23_0 defaults\n", "llvmlite 0.36.0 py38h612dafd_4 defaults\n", "locket 0.2.1 py38h06a4308_1 defaults\n", "lockfile 0.12.2 pypi_0 pypi\n", "lxml 4.6.3 pypi_0 pypi\n", "lz4-c 1.9.3 h295c915_1 defaults\n", "magics 1.5.6 pypi_0 pypi\n", "mako 1.1.4 pyh44b312d_0 conda-forge\n", "markdown 3.3.4 py38h06a4308_0 defaults\n", "markupsafe 2.0.1 pypi_0 pypi\n", "marshmallow 3.13.0 pypi_0 pypi\n", "matplotlib-base 3.4.2 py38hab158f2_0 defaults\n", "mistune 0.8.4 py38h497a2fe_1003 conda-forge\n", "mkl 2020.2 256 defaults\n", "mkl-service 2.3.0 py38he904b0f_0 defaults\n", "mkl_fft 1.3.0 py38h54f3939_0 defaults\n", "mkl_random 1.1.1 py38h0573a6f_0 defaults\n", "msgpack-python 1.0.2 py38hff7bd54_1 defaults\n", "multidict 5.1.0 py38h27cfd23_2 defaults\n", "munkres 1.1.4 py_0 defaults\n", "mypy-extensions 0.4.3 pypi_0 pypi\n", "nbclient 0.5.0 pypi_0 pypi\n", "nbconvert 6.0.7 py38h578d9bd_3 conda-forge\n", "nbdime 2.1.0 pypi_0 pypi\n", "nbformat 5.1.2 pyhd8ed1ab_1 conda-forge\n", "nbresuse 0.4.0 pypi_0 pypi\n", "nc-time-axis 1.3.1 pyhd8ed1ab_2 conda-forge\n", "ncurses 6.2 h58526e2_4 conda-forge\n", "ndg-httpsclient 0.5.1 pypi_0 pypi\n", "nest-asyncio 1.4.3 pyhd8ed1ab_0 conda-forge\n", "netcdf4 1.5.4 pypi_0 pypi\n", "networkx 2.6.3 pypi_0 pypi\n", "ninja 1.10.2 hff7bd54_1 defaults\n", "nodejs 15.3.0 h25f6087_0 conda-forge\n", "notebook 6.2.0 py38h578d9bd_0 conda-forge\n", "numba 0.53.1 py38ha9443f7_0 defaults\n", "numcodecs 0.8.0 py38h2531618_0 defaults\n", "numexpr 2.7.3 py38hb2eb853_0 defaults\n", "numpy 1.19.2 py38h54aff64_0 defaults\n", "numpy-base 1.19.2 py38hfa32c7d_0 defaults\n", "oauthlib 3.0.1 py_0 conda-forge\n", "olefile 0.46 pyhd3eb1b0_0 defaults\n", "openjpeg 2.4.0 h3ad879b_0 defaults\n", "openssl 1.1.1l h7f8727e_0 defaults\n", "opt_einsum 3.3.0 pyhd3eb1b0_1 defaults\n", "owlrl 5.2.3 pypi_0 pypi\n", "packaging 20.8 pyhd3deb0d_0 conda-forge\n", "pamela 1.0.0 py_0 conda-forge\n", "pandas 1.3.2 py38h8c16a72_0 defaults\n", "pandoc 2.11.3.2 h7f98852_0 conda-forge\n", "pandocfilters 1.4.2 py_1 conda-forge\n", "papermill 2.3.1 pypi_0 pypi\n", "parso 0.7.1 pyh9f0ad1d_0 conda-forge\n", "partd 1.2.0 pyhd3eb1b0_0 defaults\n", "pathspec 0.9.0 pypi_0 pypi\n", "patool 1.12 pypi_0 pypi\n", "pdbufr 0.9.0 pypi_0 pypi\n", "pexpect 4.8.0 pyh9f0ad1d_2 conda-forge\n", "pickleshare 0.7.5 py_1003 conda-forge\n", "pillow 8.3.1 py38h2c7a002_0 defaults\n", "pip 21.0.1 pypi_0 pypi\n", "pipx 0.16.1.0 pypi_0 pypi\n", "pluggy 0.13.1 pypi_0 pypi\n", "portalocker 2.3.2 pypi_0 pypi\n", "powerline-shell 0.7.0 pypi_0 pypi\n", "prometheus_client 0.9.0 pyhd3deb0d_0 conda-forge\n", "prompt-toolkit 3.0.10 pyha770c72_0 conda-forge\n", "properscoring 0.1 py_0 conda-forge\n", "protobuf 3.17.2 py38h295c915_0 defaults\n", "prov 1.5.1 pypi_0 pypi\n", "psutil 5.8.0 py38h27cfd23_1 defaults\n", "ptyprocess 0.7.0 pyhd3deb0d_0 conda-forge\n", "pyasn1 0.4.8 pyhd3eb1b0_0 defaults\n", "pyasn1-modules 0.2.8 py_0 defaults\n", "pycosat 0.6.3 py38h497a2fe_1006 conda-forge\n", "pycparser 2.20 pyh9f0ad1d_2 conda-forge\n", "pycurl 7.43.0.6 py38h996a351_1 conda-forge\n", "pydap 3.2.2 pyh9f0ad1d_1001 conda-forge\n", "pydot 1.4.2 pypi_0 pypi\n", "pygments 2.10.0 pypi_0 pypi\n", "pyjwt 2.1.0 pypi_0 pypi\n", "pyld 2.0.3 pypi_0 pypi\n", "pyodc 1.1.1 pypi_0 pypi\n", "pyopenssl 20.0.1 pyhd8ed1ab_0 conda-forge\n", "pyparsing 2.4.7 pyh9f0ad1d_0 conda-forge\n", "pyrsistent 0.17.3 py38h497a2fe_2 conda-forge\n", "pyshacl 0.17.0.post1 pypi_0 pypi\n", "pysocks 1.7.1 py38h578d9bd_3 conda-forge\n", "python 3.8.6 hffdb5ce_4_cpython conda-forge\n", "python-dateutil 2.8.1 py_0 conda-forge\n", "python-eccodes 2021.03.0 py38hb5d20a5_1 conda-forge\n", "python-editor 1.0.4 pypi_0 pypi\n", "python-flatbuffers 1.12 pyhd3eb1b0_0 defaults\n", "python-json-logger 2.0.1 pyh9f0ad1d_0 conda-forge\n", "python-snappy 0.6.0 py38h2531618_3 defaults\n", "python_abi 3.8 1_cp38 conda-forge\n", "pytorch 1.8.1 cpu_py38h60491be_0 defaults\n", "pytz 2021.1 pyhd3eb1b0_0 defaults\n", "pyyaml 5.4.1 pypi_0 pypi\n", "pyzmq 21.0.1 py38h3d7ac18_0 conda-forge\n", "rdflib 6.0.1 pypi_0 pypi\n", "rdflib-jsonld 0.5.0 pypi_0 pypi\n", "readline 8.0 he28a2e2_2 conda-forge\n", "regex 2021.4.4 pypi_0 pypi\n", "renku 0.16.2 pypi_0 pypi\n", "requests 2.24.0 pypi_0 pypi\n", "requests-oauthlib 1.3.0 py_0 defaults\n", "rich 10.3.0 pypi_0 pypi\n", "rsa 4.7.2 pyhd3eb1b0_1 defaults\n", "ruamel-yaml 0.16.5 pypi_0 pypi\n", "ruamel.yaml.clib 0.2.2 py38h497a2fe_2 conda-forge\n", "ruamel_yaml 0.15.80 py38h497a2fe_1003 conda-forge\n", "s3fs 2021.7.0 pyhd3eb1b0_0 defaults\n", "schema-salad 8.2.20210918131710 pypi_0 pypi\n", "scikit-learn 0.24.2 py38ha9443f7_0 defaults\n", "scipy 1.7.0 py38h7b17777_1 conda-forge\n", "send2trash 1.5.0 py_0 conda-forge\n", "setuptools 58.2.0 pypi_0 pypi\n", "setuptools-scm 6.0.1 pypi_0 pypi\n", "shellescape 3.8.1 pypi_0 pypi\n", "shellingham 1.4.0 pypi_0 pypi\n", "simpervisor 0.4 pypi_0 pypi\n", "six 1.16.0 pypi_0 pypi\n", "smmap 4.0.0 pypi_0 pypi\n", "snappy 1.1.8 he6710b0_0 defaults\n", "sortedcontainers 2.4.0 pyhd3eb1b0_0 defaults\n", "soupsieve 2.2.1 pyhd3eb1b0_0 defaults\n", "sqlalchemy 1.3.22 py38h497a2fe_1 conda-forge\n", "sqlite 3.34.0 h74cdb3f_0 conda-forge\n", "sysroot_linux-64 2.12 h77966d4_13 conda-forge\n", "tabulate 0.8.9 pypi_0 pypi\n", "tbb 2020.3 hfd86e86_0 defaults\n", "tblib 1.7.0 pyhd3eb1b0_0 defaults\n", "tenacity 7.0.0 pypi_0 pypi\n", "tensorboard 2.4.0 pyhc547734_0 defaults\n", "tensorboard-plugin-wit 1.6.0 py_0 defaults\n", "tensorflow 2.4.1 mkl_py38hb2083e0_0 defaults\n", "tensorflow-base 2.4.1 mkl_py38h43e0292_0 defaults\n", "tensorflow-estimator 2.6.0 pyh7b7c402_0 defaults\n", "termcolor 1.1.0 py38h06a4308_1 defaults\n", "terminado 0.9.2 py38h578d9bd_0 conda-forge\n", "testpath 0.4.4 py_0 conda-forge\n", "textwrap3 0.9.2 pypi_0 pypi\n", "threadpoolctl 2.2.0 pyh0d69192_0 defaults\n", "tini 0.18.0 h14c3975_1001 conda-forge\n", "tk 8.6.10 h21135ba_1 conda-forge\n", "toml 0.10.2 pypi_0 pypi\n", "toolz 0.11.1 pyhd3eb1b0_0 defaults\n", "tornado 6.1 py38h497a2fe_1 conda-forge\n", "tqdm 4.60.0 pypi_0 pypi\n", "traitlets 5.0.5 py_0 conda-forge\n", "typed-ast 1.4.2 pypi_0 pypi\n", "typing-extensions 3.7.4.3 pypi_0 pypi\n", "typing_extensions 3.10.0.2 pyh06a4308_0 defaults\n", "urllib3 1.25.11 pypi_0 pypi\n", "userpath 1.4.2 pypi_0 pypi\n", "wcmatch 8.2 pypi_0 pypi\n", "wcwidth 0.2.5 pyh9f0ad1d_2 conda-forge\n", "webencodings 0.5.1 py_1 conda-forge\n", "webob 1.8.7 pyhd3eb1b0_0 defaults\n", "werkzeug 2.0.1 pyhd3eb1b0_0 defaults\n", "wheel 0.36.2 pyhd3deb0d_0 conda-forge\n", "wrapt 1.12.1 py38h7b6447c_1 defaults\n", "xarray 0.19.0 pyhd3eb1b0_1 defaults\n", "xhistogram 0.3.0 pyhd8ed1ab_0 conda-forge\n", "xskillscore 0.0.23 pyhd8ed1ab_0 conda-forge\n", "xz 5.2.5 h516909a_1 conda-forge\n", "yagup 0.1.1 pypi_0 pypi\n", "yaml 0.2.5 h516909a_0 conda-forge\n", "yarl 1.6.3 py38h27cfd23_0 defaults\n", "zarr 2.8.1 pyhd3eb1b0_0 defaults\n", "zeromq 4.3.3 h58526e2_3 conda-forge\n", "zict 2.0.0 pyhd3eb1b0_0 defaults\n", "zipp 3.4.0 py_0 conda-forge\n", "zlib 1.2.11 h516909a_1010 conda-forge\n", "zstd 1.4.9 haebb681_0 defaults\n" ] } ], "source": [ "!conda list" ] }, { "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": "3.8.6" }, "toc-autonumbering": true }, "nbformat": 4, "nbformat_minor": 4 }