fmda/rnn_workshop.ipynb

   1 {
   2  "cells": [
   3   {
   4    "cell_type": "markdown",
   5    "id": "244c2fb0-4339-476c-a2db-a641e124e25a",
   6    "metadata": {},
   7    "source": [
   8     "# v2.1 exploration trying to make it work better"
   9    ]
  10   },
  11   {
  12    "cell_type": "code",
  13    "execution_count": null,
  14    "id": "e6cc7920-e380-4b81-bac0-cd6840450e9a",
  15    "metadata": {},
  16    "outputs": [],
  17    "source": [
  18     "# Environment\n",
  19     "import os\n",
  20     "import os.path as osp\n",
  21     "import numpy as np\n",
  22     "import pandas as pd\n",
  23     "import tensorflow as tf\n",
  24     "import matplotlib.pyplot as plt\n",
  25     "import sys\n",
  26     "# Local modules\n",
  27     "sys.path.append('..')\n",
  28     "import reproducibility\n",
  29     "import pandas as pd\n",
  30     "from utils import print_dict_summary\n",
  31     "from data_funcs import rmse, build_train_dict, combine_nested, subset_by_features\n",
  32     "from moisture_rnn import RNNParams, RNNData, RNN, RNN_LSTM, rnn_data_wrap\n",
  33     "from moisture_rnn_pkl import pkl2train\n",
  34     "from tensorflow.keras.callbacks import Callback\n",
  35     "from utils import hash2\n",
  36     "import copy\n",
  37     "import logging\n",
  38     "import pickle\n",
  39     "from utils import logging_setup, read_yml, read_pkl, hash_ndarray, hash_weights, str2time\n",
  40     "import yaml\n",
  41     "import copy\n",
  42     "import time"
  43    ]
  44   },
  45   {
  46    "cell_type": "code",
  47    "execution_count": null,
  48    "id": "f58e8839-bf0e-4995-b966-c09e4df001ce",
  49    "metadata": {},
  50    "outputs": [],
  51    "source": [
  52     "logging_setup()"
  53    ]
  54   },
  55   {
  56    "cell_type": "markdown",
  57    "id": "fae67b50-f916-45a7-bcc7-61995ba39449",
  58    "metadata": {},
  59    "source": [
  60     "## Test Data"
  61    ]
  62   },
  63   {
  64    "cell_type": "code",
  65    "execution_count": null,
  66    "id": "3efed1fa-9cda-4934-8a6c-edcf179c8755",
  67    "metadata": {},
  68    "outputs": [],
  69    "source": [
  70     "file_paths = ['data/fmda_rocky_202403-05_f05.pkl']"
  71    ]
  72   },
  73   {
  74    "cell_type": "code",
  75    "execution_count": null,
  76    "id": "28fd3746-1861-4afa-ab7e-ac449fbed322",
  77    "metadata": {},
  78    "outputs": [],
  79    "source": [
  80     "# Params used for data filtering\n",
  81     "params_data = read_yml(\"params_data.yaml\") \n",
  82     "params_data"
  83    ]
  84   },
  85   {
  86    "cell_type": "code",
  87    "execution_count": null,
  88    "id": "32a46674-8377-47f8-9c3a-e6b07f9505cf",
  89    "metadata": {},
  90    "outputs": [],
  91    "source": [
  92     "params = read_yml(\"params.yaml\", subkey='rnn') \n",
  93     "params = RNNParams(params)\n",
  94     "params.update({'epochs': 200, \n",
  95     "               'learning_rate': 0.001,\n",
  96     "               'activation': ['tanh', 'tanh'], # Activation for RNN Layers, Dense layers respectively.\n",
  97     "               'recurrent_layers': 2, 'recurrent_units': 30, \n",
  98     "               'dense_layers': 2, 'dense_units': 30,\n",
  99     "               'early_stopping_patience': 30, # how many epochs of no validation accuracy gain to wait before stopping\n",
 100     "               'batch_schedule_type': 'exp', # Hidden state batch reset schedule\n",
 101     "               'bmin': 20, # Lower bound of hidden state batch reset, \n",
 102     "               'bmax': params_data['hours'], # Upper bound of hidden state batch reset, using max hours\n",
 103     "               'features_list': ['Ed', 'Ew', 'rain', 'elev', 'lon', 'lat', 'solar', 'wind'],\n",
 104     "               'timesteps': 12\n",
 105     "              })"
 106    ]
 107   },
 108   {
 109    "cell_type": "code",
 110    "execution_count": null,
 111    "id": "c45cb8ef-41fc-4bf7-b506-dad5fd24abb3",
 112    "metadata": {},
 113    "outputs": [],
 114    "source": [
 115     "dat = read_pkl(file_paths[0])"
 116    ]
 117   },
 118   {
 119    "cell_type": "code",
 120    "execution_count": null,
 121    "id": "3c960d69-4f8a-4abb-a5d9-ed6cf98f899b",
 122    "metadata": {},
 123    "outputs": [],
 124    "source": [
 125     "import importlib\n",
 126     "import data_funcs\n",
 127     "importlib.reload(data_funcs)\n",
 128     "from data_funcs import build_train_dict"
 129    ]
 130   },
 131   {
 132    "cell_type": "code",
 133    "execution_count": null,
 134    "id": "369cd913-85cb-4855-a80c-817d84637852",
 135    "metadata": {},
 136    "outputs": [],
 137    "source": [
 138     "params_data.update({'hours': None})"
 139    ]
 140   },
 141   {
 142    "cell_type": "code",
 143    "execution_count": null,
 144    "id": "371a66ac-027d-4377-b2bb-22106707a614",
 145    "metadata": {},
 146    "outputs": [],
 147    "source": [
 148     "start_time = time.time()"
 149    ]
 150   },
 151   {
 152    "cell_type": "code",
 153    "execution_count": null,
 154    "id": "8cdc2ce8-45b4-4caa-81d9-646271ff2e97",
 155    "metadata": {
 156     "scrolled": true
 157    },
 158    "outputs": [],
 159    "source": [
 160     "train3 = build_train_dict(file_paths, params_data, spatial=False, forecast_step=3, drop_na=True)\n"
 161    ]
 162   },
 163   {
 164    "cell_type": "code",
 165    "execution_count": null,
 166    "id": "29e3289b-f47a-450d-8b17-2a7813fad089",
 167    "metadata": {},
 168    "outputs": [],
 169    "source": [
 170     "# End Timer\n",
 171     "end_time = time.time()\n",
 172     "\n",
 173     "# Calculate Code Runtime\n",
 174     "elapsed_time_sp = end_time - start_time\n",
 175     "print(f\"Spatial Training Elapsed time: {elapsed_time_sp:.4f} seconds\")"
 176    ]
 177   },
 178   {
 179    "cell_type": "code",
 180    "execution_count": null,
 181    "id": "aae373a5-f0b2-4ab9-b5df-ffa52c1bc305",
 182    "metadata": {},
 183    "outputs": [],
 184    "source": [
 185     "from data_funcs import build_features_single"
 186    ]
 187   },
 188   {
 189    "cell_type": "code",
 190    "execution_count": null,
 191    "id": "7e01fdde-ae5e-43d0-9a0d-7d7a024f4481",
 192    "metadata": {
 193     "scrolled": true
 194    },
 195    "outputs": [],
 196    "source": [
 197     "dat['PLFI1_202401'].keys()"
 198    ]
 199   },
 200   {
 201    "cell_type": "code",
 202    "execution_count": null,
 203    "id": "765867c7-72ac-4761-9d86-2558d1f75c3a",
 204    "metadata": {},
 205    "outputs": [],
 206    "source": [
 207     "start_time = time.time()\n",
 208     "\n",
 209     "for key in dat:\n",
 210     "    build_features_single(dat[key], atm=\"HRRR\", fstep=\"f03\", fprev=\"f02\")\n",
 211     "\n",
 212     "# End Timer\n",
 213     "end_time = time.time()\n",
 214     "\n",
 215     "# Calculate Code Runtime\n",
 216     "elapsed_time_sp = end_time - start_time\n",
 217     "print(f\"Spatial Training Elapsed time: {elapsed_time_sp:.4f} seconds\")"
 218    ]
 219   },
 220   {
 221    "cell_type": "code",
 222    "execution_count": null,
 223    "id": "46212800-55bd-46c6-84f9-abf395f863df",
 224    "metadata": {},
 225    "outputs": [],
 226    "source": [
 227     "from multiprocessing import Process, Queue"
 228    ]
 229   },
 230   {
 231    "cell_type": "code",
 232    "execution_count": null,
 233    "id": "cee55474-804c-4446-ab7b-ca95aa522089",
 234    "metadata": {},
 235    "outputs": [],
 236    "source": [
 237     "keys = list(dat.keys())"
 238    ]
 239   },
 240   {
 241    "cell_type": "code",
 242    "execution_count": null,
 243    "id": "bfd9aa0c-4d58-4af2-b0b2-30592fe7b4f6",
 244    "metadata": {},
 245    "outputs": [],
 246    "source": [
 247     "def process_key(key):\n",
 248     "    build_features_single(dat[key], atm=\"HRRR\", fstep=\"f03\", fprev=\"f02\")"
 249    ]
 250   },
 251   {
 252    "cell_type": "code",
 253    "execution_count": null,
 254    "id": "50836818-1961-4012-b820-9930939b3a8a",
 255    "metadata": {},
 256    "outputs": [],
 257    "source": [
 258     "from multiprocessing import Pool"
 259    ]
 260   },
 261   {
 262    "cell_type": "code",
 263    "execution_count": null,
 264    "id": "890b5fce-3dcc-47b9-8ab7-2651582ffdb5",
 265    "metadata": {},
 266    "outputs": [],
 267    "source": [
 268     "if __name__ == '__main__':\n",
 269     "    with Pool() as pool:\n",
 270     "        pool.map(process_key, keys)"
 271    ]
 272   },
 273   {
 274    "cell_type": "code",
 275    "execution_count": null,
 276    "id": "3c4548ae-caa4-4bc4-9122-9f24e7e59ef7",
 277    "metadata": {},
 278    "outputs": [],
 279    "source": []
 280   },
 281   {
 282    "cell_type": "code",
 283    "execution_count": null,
 284    "id": "3dbb6f24-4435-47b3-90c6-6176582b0d4c",
 285    "metadata": {},
 286    "outputs": [],
 287    "source": []
 288   },
 289   {
 290    "cell_type": "markdown",
 291    "id": "6322f0bc-107d-40a5-96dc-804495085a99",
 292    "metadata": {
 293     "jp-MarkdownHeadingCollapsed": true
 294    },
 295    "source": [
 296     "## Test Other ML"
 297    ]
 298   },
 299   {
 300    "cell_type": "code",
 301    "execution_count": null,
 302    "id": "12992b9a-407f-4131-ac61-e1dc338386bf",
 303    "metadata": {},
 304    "outputs": [],
 305    "source": [
 306     "params = read_yml(\"params.yaml\", subkey='xgb')\n",
 307     "params"
 308    ]
 309   },
 310   {
 311    "cell_type": "code",
 312    "execution_count": null,
 313    "id": "f214fdf8-bb76-4912-8f8c-5d0c8c1230c2",
 314    "metadata": {},
 315    "outputs": [],
 316    "source": [
 317     "dat = read_pkl(\"data/train.pkl\")"
 318    ]
 319   },
 320   {
 321    "cell_type": "code",
 322    "execution_count": null,
 323    "id": "888b7805-15f6-4c09-a05b-7aed7d253f6e",
 324    "metadata": {},
 325    "outputs": [],
 326    "source": [
 327     "cases = [*dat.keys()]"
 328    ]
 329   },
 330   {
 331    "cell_type": "code",
 332    "execution_count": null,
 333    "id": "375055d8-c070-4639-9561-e47d3f21f1f8",
 334    "metadata": {},
 335    "outputs": [],
 336    "source": [
 337     "rnn_dat = RNNData(dat[cases[10]], params['scaler'], params['features_list'])\n",
 338     "rnn_dat.train_test_split(\n",
 339     "    time_fracs = [.8, .1, .1]\n",
 340     ")\n",
 341     "rnn_dat.scale_data()"
 342    ]
 343   },
 344   {
 345    "cell_type": "code",
 346    "execution_count": null,
 347    "id": "e79f8dc8-5cf8-4190-b4ff-e640f61bd78b",
 348    "metadata": {},
 349    "outputs": [],
 350    "source": [
 351     "from moisture_models import XGB, RF, LM"
 352    ]
 353   },
 354   {
 355    "cell_type": "code",
 356    "execution_count": null,
 357    "id": "b3aeb47f-261e-4e29-9eeb-67215e5628f6",
 358    "metadata": {},
 359    "outputs": [],
 360    "source": [
 361     "mod = XGB(params)"
 362    ]
 363   },
 364   {
 365    "cell_type": "code",
 366    "execution_count": null,
 367    "id": "cae9a20d-1caf-45aa-a9c4-aef21b65d9c8",
 368    "metadata": {},
 369    "outputs": [],
 370    "source": [
 371     "mod.params"
 372    ]
 373   },
 374   {
 375    "cell_type": "code",
 376    "execution_count": null,
 377    "id": "68a07b25-c586-4fc4-a3d5-c857354e7a2c",
 378    "metadata": {},
 379    "outputs": [],
 380    "source": [
 381     "mod.fit(rnn_dat.X_train, rnn_dat.y_train)"
 382    ]
 383   },
 384   {
 385    "cell_type": "code",
 386    "execution_count": null,
 387    "id": "c8f88819-0a7a-4420-abb9-56a47015a4de",
 388    "metadata": {},
 389    "outputs": [],
 390    "source": [
 391     "preds = mod.predict(rnn_dat.X_test)"
 392    ]
 393   },
 394   {
 395    "cell_type": "code",
 396    "execution_count": null,
 397    "id": "cb7cdf14-74d6-45e4-bc1b-7d4d47dd41ac",
 398    "metadata": {},
 399    "outputs": [],
 400    "source": [
 401     "rmse(preds, rnn_dat.y_test)"
 402    ]
 403   },
 404   {
 405    "cell_type": "code",
 406    "execution_count": null,
 407    "id": "74d478c7-8c01-448e-9a00-dd0e1ee8e325",
 408    "metadata": {},
 409    "outputs": [],
 410    "source": [
 411     "plt.plot(rnn_dat.y_test)\n",
 412     "plt.plot(preds)"
 413    ]
 414   },
 415   {
 416    "cell_type": "code",
 417    "execution_count": null,
 418    "id": "c5441014-c39a-4414-a779-95b81e1ed6a8",
 419    "metadata": {},
 420    "outputs": [],
 421    "source": [
 422     "params = read_yml(\"params.yaml\", subkey='rf')\n",
 423     "rnn_dat = RNNData(dat[cases[10]], features_list = ['Ed', 'Ew', 'solar', 'wind', 'rain'])\n",
 424     "rnn_dat.train_test_split(\n",
 425     "    time_fracs = [.8, .1, .1]\n",
 426     ")"
 427    ]
 428   },
 429   {
 430    "cell_type": "code",
 431    "execution_count": null,
 432    "id": "cafe711a-20cb-4bd3-a4bc-4995a843a021",
 433    "metadata": {},
 434    "outputs": [],
 435    "source": [
 436     "import importlib\n",
 437     "import moisture_models\n",
 438     "importlib.reload(moisture_models)"
 439    ]
 440   },
 441   {
 442    "cell_type": "code",
 443    "execution_count": null,
 444    "id": "ee45f7d6-f57f-4ff6-995a-527565565f94",
 445    "metadata": {},
 446    "outputs": [],
 447    "source": [
 448     "params"
 449    ]
 450   },
 451   {
 452    "cell_type": "code",
 453    "execution_count": null,
 454    "id": "fafe76e5-0212-4bd1-a058-535935a08780",
 455    "metadata": {},
 456    "outputs": [],
 457    "source": [
 458     "mod2 = RF(params)\n",
 459     "mod2.fit(rnn_dat.X_train, rnn_dat.y_train.flatten())\n",
 460     "preds2 = mod2.predict(rnn_dat.X_test)\n",
 461     "print(rmse(preds2, rnn_dat.y_test.flatten()))\n",
 462     "plt.plot(rnn_dat.y_test)\n",
 463     "plt.plot(preds2)"
 464    ]
 465   },
 466   {
 467    "cell_type": "code",
 468    "execution_count": null,
 469    "id": "c0ab4244-996c-49af-bf4a-8b0c47b0b6db",
 470    "metadata": {},
 471    "outputs": [],
 472    "source": [
 473     "from moisture_models import RF\n",
 474     "mod2 = RF(params)"
 475    ]
 476   },
 477   {
 478    "cell_type": "code",
 479    "execution_count": null,
 480    "id": "aa6c33fd-db35-4c77-9eee-fdb39a934959",
 481    "metadata": {},
 482    "outputs": [],
 483    "source": []
 484   },
 485   {
 486    "cell_type": "code",
 487    "execution_count": null,
 488    "id": "c5598bfe-2d87-4d23-869e-aff127782462",
 489    "metadata": {},
 490    "outputs": [],
 491    "source": [
 492     "params = read_yml(\"params.yaml\", subkey='lm')\n",
 493     "rnn_dat = RNNData(dat[cases[10]], features_list = ['Ed', 'Ew', 'solar', 'wind', 'rain'])\n",
 494     "rnn_dat.train_test_split(\n",
 495     "    time_fracs = [.8, .1, .1]\n",
 496     ")\n",
 497     "mod = LM(params)"
 498    ]
 499   },
 500   {
 501    "cell_type": "code",
 502    "execution_count": null,
 503    "id": "d828c15c-4078-4967-abff-c1fd15d4696d",
 504    "metadata": {},
 505    "outputs": [],
 506    "source": [
 507     "mod.fit(rnn_dat.X_train, rnn_dat.y_train)\n",
 508     "preds = mod.predict(rnn_dat.X_test)\n",
 509     "print(rmse(preds2, rnn_dat.y_test.flatten()))"
 510    ]
 511   },
 512   {
 513    "cell_type": "code",
 514    "execution_count": null,
 515    "id": "8496a32a-8269-4d6b-953e-7f33fe626789",
 516    "metadata": {},
 517    "outputs": [],
 518    "source": []
 519   },
 520   {
 521    "cell_type": "code",
 522    "execution_count": null,
 523    "id": "75ce8bf3-6efb-4dc7-b895-def92f6ce6b4",
 524    "metadata": {},
 525    "outputs": [],
 526    "source": []
 527   },
 528   {
 529    "cell_type": "markdown",
 530    "id": "282cb651-b21f-401d-94c5-9e07530a9ba8",
 531    "metadata": {},
 532    "source": [
 533     "## RNN"
 534    ]
 535   },
 536   {
 537    "cell_type": "code",
 538    "execution_count": null,
 539    "id": "fa38f35a-d367-4df8-b2d3-7691ff4b0cf4",
 540    "metadata": {},
 541    "outputs": [],
 542    "source": []
 543   },
 544   {
 545    "cell_type": "markdown",
 546    "id": "5ef092ff-8af1-491a-b0bf-cc3e674330e0",
 547    "metadata": {},
 548    "source": [
 549     "## Phys Initialized"
 550    ]
 551   },
 552   {
 553    "cell_type": "code",
 554    "execution_count": null,
 555    "id": "5488628e-4552-4909-83e9-413fd6878bdd",
 556    "metadata": {},
 557    "outputs": [],
 558    "source": [
 559     "params.update({\n",
 560     "    'epochs':100,\n",
 561     "    'dense_layers': 0,\n",
 562     "    'activation': ['relu', 'relu'],\n",
 563     "    'phys_initialize': False,\n",
 564     "    'dropout': [0,0],\n",
 565     "    'space_fracs': [.8, .1, .1],\n",
 566     "    'scaler': None\n",
 567     "})"
 568    ]
 569   },
 570   {
 571    "cell_type": "code",
 572    "execution_count": null,
 573    "id": "ab7db7d6-949e-457d-90b9-22d9c5aa4739",
 574    "metadata": {},
 575    "outputs": [],
 576    "source": [
 577     "import importlib\n",
 578     "import moisture_rnn\n",
 579     "importlib.reload(moisture_rnn)\n",
 580     "from moisture_rnn import rnn_data_wrap"
 581    ]
 582   },
 583   {
 584    "cell_type": "code",
 585    "execution_count": null,
 586    "id": "d26cf1b2-2fad-409d-888f-4921b0ae4ba8",
 587    "metadata": {},
 588    "outputs": [],
 589    "source": [
 590     "params['scaler'] is None"
 591    ]
 592   },
 593   {
 594    "cell_type": "code",
 595    "execution_count": null,
 596    "id": "1c4627bc-0f90-44e6-9103-2efe5c5f439d",
 597    "metadata": {},
 598    "outputs": [],
 599    "source": [
 600     "rnn_dat = rnn_data_wrap(combine_nested(train3), params)"
 601    ]
 602   },
 603   {
 604    "cell_type": "code",
 605    "execution_count": null,
 606    "id": "56bdf26c-07e7-4e4a-a567-af7dd0f564d9",
 607    "metadata": {},
 608    "outputs": [],
 609    "source": [
 610     "reproducibility.set_seed()\n",
 611     "rnn = RNN(params)\n",
 612     "m, errs = rnn.run_model(rnn_dat)"
 613    ]
 614   },
 615   {
 616    "cell_type": "code",
 617    "execution_count": null,
 618    "id": "01227b79-98f3-4931-bdfc-ff08afa8be5f",
 619    "metadata": {},
 620    "outputs": [],
 621    "source": [
 622     "rnn.model_train.summary()"
 623    ]
 624   },
 625   {
 626    "cell_type": "code",
 627    "execution_count": null,
 628    "id": "918a8bf0-638b-4b4b-82fe-c6a1965a72dd",
 629    "metadata": {},
 630    "outputs": [],
 631    "source": [
 632     "errs.mean()"
 633    ]
 634   },
 635   {
 636    "cell_type": "code",
 637    "execution_count": null,
 638    "id": "37fdbb3a-3e83-4541-93b2-982b6d4cbe93",
 639    "metadata": {
 640     "scrolled": true
 641    },
 642    "outputs": [],
 643    "source": [
 644     "rnn_dat.X_train[:,:,0].mean()"
 645    ]
 646   },
 647   {
 648    "cell_type": "code",
 649    "execution_count": null,
 650    "id": "7ca41db1-72aa-44b6-b9dd-058735336ab3",
 651    "metadata": {},
 652    "outputs": [],
 653    "source": []
 654   },
 655   {
 656    "cell_type": "code",
 657    "execution_count": null,
 658    "id": "a592a4c9-cb3b-4174-8eaa-02afd00a1897",
 659    "metadata": {},
 660    "outputs": [],
 661    "source": [
 662     "rnn_dat['features_list']"
 663    ]
 664   },
 665   {
 666    "cell_type": "code",
 667    "execution_count": null,
 668    "id": "3832fb05-417c-4648-8e2e-7748c06b3768",
 669    "metadata": {},
 670    "outputs": [],
 671    "source": []
 672   },
 673   {
 674    "cell_type": "markdown",
 675    "id": "d2360aef-e9c4-4a71-922d-336e53b82537",
 676    "metadata": {},
 677    "source": [
 678     "## LSTM"
 679    ]
 680   },
 681   {
 682    "cell_type": "code",
 683    "execution_count": null,
 684    "id": "71d4e441-9bf1-4d57-bb37-091553e23212",
 685    "metadata": {},
 686    "outputs": [],
 687    "source": [
 688     "import importlib \n",
 689     "import moisture_rnn\n",
 690     "importlib.reload(moisture_rnn)\n",
 691     "from moisture_rnn import RNN_LSTM"
 692    ]
 693   },
 694   {
 695    "cell_type": "code",
 696    "execution_count": null,
 697    "id": "0f6ba896-e3be-4a9f-8a42-3df64aff7d63",
 698    "metadata": {},
 699    "outputs": [],
 700    "source": [
 701     "params = read_yml(\"params.yaml\", subkey=\"lstm\")\n",
 702     "params = RNNParams(params)"
 703    ]
 704   },
 705   {
 706    "cell_type": "code",
 707    "execution_count": null,
 708    "id": "a4cf567e-d623-4e14-b578-eed88b80d04e",
 709    "metadata": {},
 710    "outputs": [],
 711    "source": [
 712     "rnn_dat = rnn_data_wrap(combine_nested(train3), params)"
 713    ]
 714   },
 715   {
 716    "cell_type": "code",
 717    "execution_count": null,
 718    "id": "57bb5708-7be9-4474-abb4-3b7ff4bf79df",
 719    "metadata": {},
 720    "outputs": [],
 721    "source": [
 722     "params.update({\n",
 723     "    'loc_batch_reset': rnn_dat.n_seqs # Used to reset hidden state when location changes for a given batch\n",
 724     "})"
 725    ]
 726   },
 727   {
 728    "cell_type": "code",
 729    "execution_count": null,
 730    "id": "0157a6bc-3a99-4b87-a42c-ab770d19ae37",
 731    "metadata": {},
 732    "outputs": [],
 733    "source": [
 734     "from moisture_rnn import ResetStatesCallback, EarlyStoppingCallback\n",
 735     "params.update({'epochs': 50, 'learning_rate': 0.001, 'verbose_fit': True, 'rnn_layers': 2, 'rnn_units': 20, 'dense_layers': 1, 'dense_units': 10,\n",
 736     "              'activation': ['tanh', 'tanh'], 'features_list': rnn_dat.features_list,\n",
 737     "              'batch_schedule_type':'exp', 'bmin': 10, 'bmax':rnn_dat.hours})\n",
 738     "reproducibility.set_seed(123)\n",
 739     "lstm = RNN_LSTM(params)\n",
 740     "\n",
 741     "history = lstm.model_train.fit(rnn_dat.X_train, rnn_dat.y_train, \n",
 742     "                    batch_size = params['batch_size'], epochs=params['epochs'], \n",
 743     "                    callbacks = [ResetStatesCallback(params),\n",
 744     "                                EarlyStoppingCallback(patience = 15)],\n",
 745     "                   validation_data = (rnn_dat.X_val, rnn_dat.y_val))\n",
 746     "              "
 747    ]
 748   },
 749   {
 750    "cell_type": "code",
 751    "execution_count": null,
 752    "id": "ec95e7d4-6d57-441b-b673-f10625ee5dec",
 753    "metadata": {},
 754    "outputs": [],
 755    "source": [
 756     "errs.mean()"
 757    ]
 758   },
 759   {
 760    "cell_type": "code",
 761    "execution_count": null,
 762    "id": "9b3c8d8d-ea50-44ea-8c0c-414e07cd01ac",
 763    "metadata": {},
 764    "outputs": [],
 765    "source": []
 766   },
 767   {
 768    "cell_type": "code",
 769    "execution_count": null,
 770    "id": "03063e3c-e8f4-451d-b0cf-25bd965cd9d6",
 771    "metadata": {},
 772    "outputs": [],
 773    "source": [
 774     "params = RNNParams(read_yml(\"params.yaml\", subkey=\"lstm\"))\n",
 775     "params.update({'epochs': 50, 'learning_rate': 0.001, 'verbose_fit': True, 'rnn_layers': 2, 'rnn_units': 20, 'dense_layers': 1, 'dense_units': 10,\n",
 776     "              'activation': ['tanh', 'tanh'], 'features_list': rnn_dat.features_list,\n",
 777     "              'batch_schedule_type':'exp', 'bmin': 10, 'bmax':rnn_dat.hours,\n",
 778     "              'early_stopping_patience': 25})\n",
 779     "rnn_dat = rnn_data_wrap(combine_nested(train3), params)\n",
 780     "params.update({\n",
 781     "    'loc_batch_reset': rnn_dat.n_seqs # Used to reset hidden state when location changes for a given batch\n",
 782     "})\n",
 783     "reproducibility.set_seed(123)\n",
 784     "lstm = RNN_LSTM(params)\n",
 785     "m, errs = lstm.run_model(rnn_dat)"
 786    ]
 787   },
 788   {
 789    "cell_type": "code",
 790    "execution_count": null,
 791    "id": "be46a2dc-bf5c-4893-a1ee-a1682566f7a2",
 792    "metadata": {},
 793    "outputs": [],
 794    "source": [
 795     "errs.mean()"
 796    ]
 797   },
 798   {
 799    "cell_type": "code",
 800    "execution_count": null,
 801    "id": "0f319f37-7d13-41fd-95fa-66dbdfeab588",
 802    "metadata": {},
 803    "outputs": [],
 804    "source": []
 805   },
 806   {
 807    "cell_type": "code",
 808    "execution_count": null,
 809    "id": "b1252b08-62b9-4d24-add2-0f87d15b0ff2",
 810    "metadata": {},
 811    "outputs": [],
 812    "source": [
 813     "params = RNNParams(read_yml(\"params.yaml\", subkey=\"rnn\"))\n",
 814     "rnn_dat = rnn_data_wrap(combine_nested(train3), params)"
 815    ]
 816   },
 817   {
 818    "cell_type": "code",
 819    "execution_count": null,
 820    "id": "9281540b-eb26-4923-883b-1b31d8347634",
 821    "metadata": {},
 822    "outputs": [],
 823    "source": [
 824     "reproducibility.set_seed(123)\n",
 825     "rnn = RNN(params)\n",
 826     "m, errs = rnn.run_model(rnn_dat)"
 827    ]
 828   },
 829   {
 830    "cell_type": "code",
 831    "execution_count": null,
 832    "id": "8a0269b4-d6b7-4f20-8386-69814d7acaa3",
 833    "metadata": {},
 834    "outputs": [],
 835    "source": [
 836     "errs.mean()"
 837    ]
 838   },
 839   {
 840    "cell_type": "code",
 841    "execution_count": null,
 842    "id": "10b44de3-a0e9-49e4-9e03-873d69580c07",
 843    "metadata": {},
 844    "outputs": [],
 845    "source": []
 846   },
 847   {
 848    "cell_type": "code",
 849    "execution_count": null,
 850    "id": "27f4fee4-7fce-49c5-a455-97a90b754c13",
 851    "metadata": {},
 852    "outputs": [],
 853    "source": []
 854   },
 855   {
 856    "cell_type": "code",
 857    "execution_count": null,
 858    "id": "739d4b26-641e-47b2-a90a-67cd32215d05",
 859    "metadata": {},
 860    "outputs": [],
 861    "source": []
 862   }
 863  ],
 864  "metadata": {
 865   "kernelspec": {
 866    "display_name": "Python 3 (ipykernel)",
 867    "language": "python",
 868    "name": "python3"
 869   },
 870   "language_info": {
 871    "codemirror_mode": {
 872     "name": "ipython",
 873     "version": 3
 874    },
 875    "file_extension": ".py",
 876    "mimetype": "text/x-python",
 877    "name": "python",
 878    "nbconvert_exporter": "python",
 879    "pygments_lexer": "ipython3",
 880    "version": "3.12.5"
 881   }
 882  },
 883  "nbformat": 4,
 884  "nbformat_minor": 5
 885 }