reports/report_2017-03-01.7h_average.ipynb

   1 {
   2  "cells": [
   3   {
   4    "cell_type": "code",
   5    "execution_count": null,
   6    "metadata": {
   7     "collapsed": false
   8    },
   9    "outputs": [],
  10    "source": [
  11     "library(talweg)"
  12    ]
  13   },
  14   {
  15    "cell_type": "code",
  16    "execution_count": null,
  17    "metadata": {
  18     "collapsed": false
  19    },
  20    "outputs": [],
  21    "source": [
  22     "ts_data = read.csv(system.file(\"extdata\",\"pm10_mesures_H_loc.csv\",package=\"talweg\"))\n",
  23     "exo_data = read.csv(system.file(\"extdata\",\"meteo_extra_noNAs.csv\",package=\"talweg\"))\n",
  24     "data = getData(ts_data, exo_data, input_tz = \"Europe/Paris\", working_tz=\"Europe/Paris\", predict_at=7)"
  25    ]
  26   },
  27   {
  28    "cell_type": "markdown",
  29    "metadata": {},
  30    "source": [
  31     "## Introduction\n",
  32     "\n",
  33     "J'ai fait quelques essais dans différentes configurations pour la méthode \"Neighbors\" (la seule dont on a parlé).<br>Il semble que le mieux soit\n",
  34     "\n",
  35     " * simtype=\"mix\" : on utilise les similarités endogènes et exogènes (fenêtre optimisée par VC)\n",
  36     " * same_season=FALSE : les indices pour la validation croisée ne tiennent pas compte des saisons\n",
  37     " * mix_strategy=\"mult\" : on multiplie les poids (au lieu d'en éteindre)\n",
  38     "\n",
  39     "(valeurs par défaut).\n",
  40     "\n",
  41     "J'ai systématiquement comparé à deux autres approches : la persistence et la moyennes des lendemains des jours \"similaires\" dans tout le passé ; à chaque fois sans prédiction du saut (sauf pour Neighbors : prédiction basée sur les poids calculés).\n",
  42     "\n",
  43     "Ensuite j'affiche les erreurs, quelques courbes prévues/mesurées, quelques filaments puis les histogrammes de quelques poids. Concernant les graphes de filaments, la moitié gauche du graphe correspond aux jours similaires au jour courant, tandis que la moitié droite affiche les lendemains : ce sont donc les voisinages tels qu'utilisés dans l'algorithme.\n",
  44     "\n",
  45     "<h2 style=\"color:blue;font-size:2em\">Pollution par chauffage</h2>"
  46    ]
  47   },
  48   {
  49    "cell_type": "code",
  50    "execution_count": null,
  51    "metadata": {
  52     "collapsed": false
  53    },
  54    "outputs": [],
  55    "source": [
  56     "indices_ch = seq(as.Date(\"2015-01-18\"),as.Date(\"2015-01-24\"),\"days\")\n",
  57     "p_ch_nn = computeForecast(data, indices_ch, \"Neighbors\", \"Neighbors\", simtype=\"mix\")\n",
  58     "p_ch_pz = computeForecast(data, indices_ch, \"Persistence\", \"Zero\", same_day=TRUE)\n",
  59     "p_ch_az = computeForecast(data, indices_ch, \"Average\", \"Zero\") #, memory=183)\n",
  60     "#p_ch_zz = computeForecast(data, indices_ch, \"Zero\", \"Zero\")"
  61    ]
  62   },
  63   {
  64    "cell_type": "code",
  65    "execution_count": null,
  66    "metadata": {
  67     "collapsed": false
  68    },
  69    "outputs": [],
  70    "source": [
  71     "e_ch_nn = computeError(data, p_ch_nn)\n",
  72     "e_ch_pz = computeError(data, p_ch_pz)\n",
  73     "e_ch_az = computeError(data, p_ch_az)\n",
  74     "#e_ch_zz = computeError(data, p_ch_zz)\n",
  75     "options(repr.plot.width=9, repr.plot.height=7)\n",
  76     "plotError(list(e_ch_nn, e_ch_pz, e_ch_az), cols=c(1,2,colors()[258]))\n",
  77     "\n",
  78     "#Noir: neighbors, rouge: persistence, vert: moyenne"
  79    ]
  80   },
  81   {
  82    "cell_type": "code",
  83    "execution_count": null,
  84    "metadata": {
  85     "collapsed": false
  86    },
  87    "outputs": [],
  88    "source": [
  89     "par(mfrow=c(1,2))\n",
  90     "options(repr.plot.width=9, repr.plot.height=4)\n",
  91     "plotPredReal(data, p_ch_nn, 3)\n",
  92     "plotPredReal(data, p_ch_nn, 4)\n",
  93     "\n",
  94     "#Bleu: prévue, noir: réalisée"
  95    ]
  96   },
  97   {
  98    "cell_type": "code",
  99    "execution_count": null,
 100    "metadata": {
 101     "collapsed": false
 102    },
 103    "outputs": [],
 104    "source": [
 105     "par(mfrow=c(1,2))\n",
 106     "plotPredReal(data, p_ch_az, 3)\n",
 107     "plotPredReal(data, p_ch_az, 4)"
 108    ]
 109   },
 110   {
 111    "cell_type": "code",
 112    "execution_count": null,
 113    "metadata": {
 114     "collapsed": false
 115    },
 116    "outputs": [],
 117    "source": [
 118     "par(mfrow=c(1,2))\n",
 119     "f3_ch = computeFilaments(data, p_ch_nn$getIndexInData(3), plot=TRUE)\n",
 120     "f4_ch = computeFilaments(data, p_ch_nn$getIndexInData(4), plot=TRUE)"
 121    ]
 122   },
 123   {
 124    "cell_type": "code",
 125    "execution_count": null,
 126    "metadata": {
 127     "collapsed": false
 128    },
 129    "outputs": [],
 130    "source": [
 131     "par(mfrow=c(1,2))\n",
 132     "plotFilamentsBox(data, f3_ch)\n",
 133     "plotFilamentsBox(data, f4_ch)\n",
 134     "\n",
 135     "#À gauche : jour 3 + lendemain (4) ; à droite : jour 4 + lendemain (5)"
 136    ]
 137   },
 138   {
 139    "cell_type": "code",
 140    "execution_count": null,
 141    "metadata": {
 142     "collapsed": false
 143    },
 144    "outputs": [],
 145    "source": [
 146     "par(mfrow=c(1,2))\n",
 147     "plotRelVar(data, f3_ch)\n",
 148     "plotRelVar(data, f4_ch)\n",
 149     "\n",
 150     "#Variabilité globale en rouge ; sur nos 60 voisins (+ lendemains) en noir"
 151    ]
 152   },
 153   {
 154    "cell_type": "code",
 155    "execution_count": null,
 156    "metadata": {
 157     "collapsed": false
 158    },
 159    "outputs": [],
 160    "source": [
 161     "par(mfrow=c(1,2))\n",
 162     "plotSimils(p_ch_nn, 3)\n",
 163     "plotSimils(p_ch_nn, 4)\n",
 164     "\n",
 165     "#Non pollué à gauche, pollué à droite"
 166    ]
 167   },
 168   {
 169    "cell_type": "code",
 170    "execution_count": null,
 171    "metadata": {
 172     "collapsed": false
 173    },
 174    "outputs": [],
 175    "source": [
 176     "#Fenêtres sélectionnées dans ]0,10] / endo à gauche, exo à droite\n",
 177     "p_ch_nn$getParams(3)$window\n",
 178     "p_ch_nn$getParams(4)$window"
 179    ]
 180   },
 181   {
 182    "cell_type": "markdown",
 183    "metadata": {},
 184    "source": [
 185     "<h2 style=\"color:blue;font-size:2em\">Pollution par épandage</h2>"
 186    ]
 187   },
 188   {
 189    "cell_type": "code",
 190    "execution_count": null,
 191    "metadata": {
 192     "collapsed": false
 193    },
 194    "outputs": [],
 195    "source": [
 196     "indices_ep = seq(as.Date(\"2015-03-15\"),as.Date(\"2015-03-21\"),\"days\")\n",
 197     "p_ep_nn = computeForecast(data,indices_ep, \"Neighbors\", \"Neighbors\", simtype=\"mix\")\n",
 198     "p_ep_pz = computeForecast(data, indices_ep, \"Persistence\", \"Zero\", same_day=TRUE)\n",
 199     "p_ep_az = computeForecast(data, indices_ep, \"Average\", \"Zero\") #, memory=183)\n",
 200     "#p_ep_zz = computeForecast(data, indices_ep, \"Zero\", \"Zero\")"
 201    ]
 202   },
 203   {
 204    "cell_type": "code",
 205    "execution_count": null,
 206    "metadata": {
 207     "collapsed": false
 208    },
 209    "outputs": [],
 210    "source": [
 211     "e_ep_nn = computeError(data, p_ep_nn)\n",
 212     "e_ep_pz = computeError(data, p_ep_pz)\n",
 213     "e_ep_az = computeError(data, p_ep_az)\n",
 214     "#e_ep_zz = computeError(data, p_ep_zz)\n",
 215     "options(repr.plot.width=9, repr.plot.height=7)\n",
 216     "plotError(list(e_ep_nn, e_ep_pz, e_ep_az), cols=c(1,2,colors()[258]))\n",
 217     "\n",
 218     "#Noir: neighbors, rouge: persistence, vert: moyenne"
 219    ]
 220   },
 221   {
 222    "cell_type": "code",
 223    "execution_count": null,
 224    "metadata": {
 225     "collapsed": false
 226    },
 227    "outputs": [],
 228    "source": [
 229     "par(mfrow=c(1,2))\n",
 230     "options(repr.plot.width=9, repr.plot.height=4)\n",
 231     "plotPredReal(data, p_ep_nn, 4)\n",
 232     "plotPredReal(data, p_ep_nn, 6)\n",
 233     "\n",
 234     "#Bleu: prévue, noir: réalisée"
 235    ]
 236   },
 237   {
 238    "cell_type": "code",
 239    "execution_count": null,
 240    "metadata": {
 241     "collapsed": false
 242    },
 243    "outputs": [],
 244    "source": [
 245     "par(mfrow=c(1,2))\n",
 246     "plotPredReal(data, p_ep_az, 4)\n",
 247     "plotPredReal(data, p_ep_az, 6)"
 248    ]
 249   },
 250   {
 251    "cell_type": "code",
 252    "execution_count": null,
 253    "metadata": {
 254     "collapsed": false
 255    },
 256    "outputs": [],
 257    "source": [
 258     "par(mfrow=c(1,2))\n",
 259     "f4_ep = computeFilaments(data, p_ep_nn$getIndexInData(4), plot=TRUE)\n",
 260     "f6_ep = computeFilaments(data, p_ep_nn$getIndexInData(6), plot=TRUE)"
 261    ]
 262   },
 263   {
 264    "cell_type": "code",
 265    "execution_count": null,
 266    "metadata": {
 267     "collapsed": false
 268    },
 269    "outputs": [],
 270    "source": [
 271     "par(mfrow=c(2,2))\n",
 272     "plotFilamentsBox(data, f4_ep)\n",
 273     "plotFilamentsBox(data, f6_ep)\n",
 274     "\n",
 275     "#À gauche : jour 4 + lendemain (5) ; à droite : jour 6 + lendemain (7)"
 276    ]
 277   },
 278   {
 279    "cell_type": "code",
 280    "execution_count": null,
 281    "metadata": {
 282     "collapsed": false
 283    },
 284    "outputs": [],
 285    "source": [
 286     "par(mfrow=c(1,2))\n",
 287     "plotRelativeVariability(data, f4_ep)\n",
 288     "plotRelativeVariability(data, f6_ep)\n",
 289     "\n",
 290     "#Variabilité globale en rouge ; sur nos 60 voisins (+ lendemains) en noir"
 291    ]
 292   },
 293   {
 294    "cell_type": "code",
 295    "execution_count": null,
 296    "metadata": {
 297     "collapsed": false
 298    },
 299    "outputs": [],
 300    "source": [
 301     "par(mfrow=c(1,2))\n",
 302     "plotSimils(p_ep_nn, 4)\n",
 303     "plotSimils(p_ep_nn, 6)"
 304    ]
 305   },
 306   {
 307    "cell_type": "code",
 308    "execution_count": null,
 309    "metadata": {
 310     "collapsed": false
 311    },
 312    "outputs": [],
 313    "source": [
 314     "#Fenêtres sélectionnées dans ]0,10] / endo à gauche, exo à droite\n",
 315     "p_ep_nn$getParams(4)$window\n",
 316     "p_ep_nn$getParams(6)$window"
 317    ]
 318   },
 319   {
 320    "cell_type": "markdown",
 321    "metadata": {},
 322    "source": [
 323     "<h2 style=\"color:blue;font-size:2em\">Semaine non polluée</h2>"
 324    ]
 325   },
 326   {
 327    "cell_type": "code",
 328    "execution_count": null,
 329    "metadata": {
 330     "collapsed": false
 331    },
 332    "outputs": [],
 333    "source": [
 334     "indices_np = seq(as.Date(\"2015-04-26\"),as.Date(\"2015-05-02\"),\"days\")\n",
 335     "p_np_nn = computeForecast(data,indices_np, \"Neighbors\", \"Neighbors\", simtype=\"mix\")\n",
 336     "p_np_pz = computeForecast(data, indices_np, \"Persistence\", \"Zero\", same_day=FALSE)\n",
 337     "p_np_az = computeForecast(data, indices_np, \"Average\", \"Zero\") #, memory=183)\n",
 338     "#p_np_zz = computeForecast(data, indices_np, \"Zero\", \"Zero\")"
 339    ]
 340   },
 341   {
 342    "cell_type": "code",
 343    "execution_count": null,
 344    "metadata": {
 345     "collapsed": false
 346    },
 347    "outputs": [],
 348    "source": [
 349     "e_np_nn = computeError(data, p_np_nn)\n",
 350     "e_np_pz = computeError(data, p_np_pz)\n",
 351     "e_np_az = computeError(data, p_np_az)\n",
 352     "#e_np_zz = computeError(data, p_np_zz)\n",
 353     "options(repr.plot.width=9, repr.plot.height=7)\n",
 354     "plotError(list(e_np_nn, e_np_pz, e_np_az), cols=c(1,2,colors()[258]))\n",
 355     "\n",
 356     "#Noir: neighbors, rouge: persistence, vert: moyenne"
 357    ]
 358   },
 359   {
 360    "cell_type": "code",
 361    "execution_count": null,
 362    "metadata": {
 363     "collapsed": false
 364    },
 365    "outputs": [],
 366    "source": [
 367     "par(mfrow=c(1,2))\n",
 368     "options(repr.plot.width=9, repr.plot.height=4)\n",
 369     "plotPredReal(data, p_np_nn, 5)\n",
 370     "plotPredReal(data, p_np_nn, 6)\n",
 371     "\n",
 372     "#Bleu: prévue, noir: réalisée"
 373    ]
 374   },
 375   {
 376    "cell_type": "code",
 377    "execution_count": null,
 378    "metadata": {
 379     "collapsed": false
 380    },
 381    "outputs": [],
 382    "source": [
 383     "par(mfrow=c(1,2))\n",
 384     "plotPredReal(data, p_np_az, 5)\n",
 385     "plotPredReal(data, p_np_az, 6)"
 386    ]
 387   },
 388   {
 389    "cell_type": "code",
 390    "execution_count": null,
 391    "metadata": {
 392     "collapsed": false
 393    },
 394    "outputs": [],
 395    "source": [
 396     "par(mfrow=c(1,2))\n",
 397     "f5_np = computeFilaments(data, p_np_nn$getIndexInData(5), plot=TRUE)\n",
 398     "f6_np = computeFilaments(data, p_np_nn$getIndexInData(6), plot=TRUE)"
 399    ]
 400   },
 401   {
 402    "cell_type": "code",
 403    "execution_count": null,
 404    "metadata": {
 405     "collapsed": false
 406    },
 407    "outputs": [],
 408    "source": [
 409     "par(mfrow=c(2,2))\n",
 410     "plotFilamentsBox(data, f5_np)\n",
 411     "plotFilamentsBox(data, f6_np)\n",
 412     "\n",
 413     "#À gauche : jour 5 + lendemain (6) ; à droite : jour 6 + lendemain (7)"
 414    ]
 415   },
 416   {
 417    "cell_type": "code",
 418    "execution_count": null,
 419    "metadata": {
 420     "collapsed": false
 421    },
 422    "outputs": [],
 423    "source": [
 424     "par(mfrow=c(1,2))\n",
 425     "plotRelVar(data, f5_np)\n",
 426     "plotRelVar(data, f6_np)\n",
 427     "\n",
 428     "#Variabilité globale en rouge ; sur nos 60 voisins (+ lendemains) en noir"
 429    ]
 430   },
 431   {
 432    "cell_type": "code",
 433    "execution_count": null,
 434    "metadata": {
 435     "collapsed": false
 436    },
 437    "outputs": [],
 438    "source": [
 439     "par(mfrow=c(1,2))\n",
 440     "plotSimils(p_np_nn, 5)\n",
 441     "plotSimils(p_np_nn, 6)"
 442    ]
 443   },
 444   {
 445    "cell_type": "code",
 446    "execution_count": null,
 447    "metadata": {
 448     "collapsed": false
 449    },
 450    "outputs": [],
 451    "source": [
 452     "#Fenêtres sélectionnées dans ]0,10] / endo à gauche, exo à droite\n",
 453     "p_np_nn$getParams(5)$window\n",
 454     "p_np_nn$getParams(6)$window"
 455    ]
 456   },
 457   {
 458    "cell_type": "markdown",
 459    "metadata": {},
 460    "source": [
 461     "## Bilan\n",
 462     "\n",
 463     "Problème difficile : on ne fait guère mieux qu'une naïve moyenne des lendemains des jours similaires dans le passé, ce qui n'est pas loin de prédire une série constante égale à la dernière valeur observée (méthode \"zéro\"). La persistence donne parfois de bons résultats mais est trop instable (sensibilité à l'argument <code>same_day</code>).\n",
 464     "\n",
 465     "Comment améliorer la méthode ?"
 466    ]
 467   }
 468  ],
 469  "metadata": {
 470   "kernelspec": {
 471    "display_name": "R",
 472    "language": "R",
 473    "name": "ir"
 474   },
 475   "language_info": {
 476    "codemirror_mode": "r",
 477    "file_extension": ".r",
 478    "mimetype": "text/x-r-source",
 479    "name": "R",
 480    "pygments_lexer": "r",
 481    "version": "3.3.2"
 482   }
 483  },
 484  "nbformat": 4,
 485  "nbformat_minor": 2
 486 }