clustering-soln.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true,
    "jupyter": {
     "outputs_hidden": true
    }
   },
   "outputs": [],
   "source": [
    "# Load libraries\n",
    "from pyspark.context import SparkContext\n",
    "from pyspark.sql import SQLContext, SparkSession\n",
    "from pyspark.ml.clustering import KMeans\n",
    "from pyspark.ml.feature import VectorAssembler\n",
    "from pyspark.ml.feature import StandardScaler\n",
    "import utils\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "# Load minute weather data\n",
    "spark = SparkSession.builder.master(\"local\").appName(\"Word Count\").getOrCreate()\n",
    "\n",
    "sc = SparkContext.getOrCreate()\n",
    "sqlContext = SQLContext(sc)\n",
    "\n",
    "inputfile = \"/srv/jupyterhub/read-only/data/minute_weather.csv\"\n",
    "df = spark.read.load (inputfile, format=\"csv\", inferSchema=\"true\", header=\"true\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- rowID: integer (nullable = true)\n",
      " |-- hpwren_timestamp: timestamp (nullable = true)\n",
      " |-- air_pressure: double (nullable = true)\n",
      " |-- air_temp: double (nullable = true)\n",
      " |-- avg_wind_direction: double (nullable = true)\n",
      " |-- avg_wind_speed: double (nullable = true)\n",
      " |-- max_wind_direction: double (nullable = true)\n",
      " |-- max_wind_speed: double (nullable = true)\n",
      " |-- min_wind_direction: double (nullable = true)\n",
      " |-- min_wind_speed: double (nullable = true)\n",
      " |-- rain_accumulation: double (nullable = true)\n",
      " |-- rain_duration: double (nullable = true)\n",
      " |-- relative_humidity: double (nullable = true)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Examine schema\n",
    "\n",
    "df.printSchema()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1587257"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Count rows\n",
    "\n",
    "df.count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "15873"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Filter rows \n",
    "\n",
    "filteredDF = df.filter((df.rowID % 100) == 0)\n",
    "filteredDF.count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>summary</th>\n",
       "      <td>count</td>\n",
       "      <td>mean</td>\n",
       "      <td>stddev</td>\n",
       "      <td>min</td>\n",
       "      <td>max</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rowID</th>\n",
       "      <td>15873</td>\n",
       "      <td>793600.0</td>\n",
       "      <td>458228.4746717515</td>\n",
       "      <td>0</td>\n",
       "      <td>1587200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>air_pressure</th>\n",
       "      <td>15873</td>\n",
       "      <td>916.8291627291593</td>\n",
       "      <td>3.0517222151797627</td>\n",
       "      <td>905.1</td>\n",
       "      <td>929.4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>air_temp</th>\n",
       "      <td>15873</td>\n",
       "      <td>61.85468909468877</td>\n",
       "      <td>11.835413790821452</td>\n",
       "      <td>32.36</td>\n",
       "      <td>96.44</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>avg_wind_direction</th>\n",
       "      <td>15870</td>\n",
       "      <td>161.2875236294896</td>\n",
       "      <td>95.31316129656483</td>\n",
       "      <td>0.0</td>\n",
       "      <td>359.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>avg_wind_speed</th>\n",
       "      <td>15870</td>\n",
       "      <td>2.792804032766223</td>\n",
       "      <td>2.07050619846002</td>\n",
       "      <td>0.1</td>\n",
       "      <td>20.1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max_wind_direction</th>\n",
       "      <td>15870</td>\n",
       "      <td>162.70094517958412</td>\n",
       "      <td>92.26960112663168</td>\n",
       "      <td>0.0</td>\n",
       "      <td>359.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max_wind_speed</th>\n",
       "      <td>15870</td>\n",
       "      <td>3.4146250787649493</td>\n",
       "      <td>2.4289064068121387</td>\n",
       "      <td>0.1</td>\n",
       "      <td>20.9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min_wind_direction</th>\n",
       "      <td>15870</td>\n",
       "      <td>166.64429741650915</td>\n",
       "      <td>97.82483630682518</td>\n",
       "      <td>0.0</td>\n",
       "      <td>359.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min_wind_speed</th>\n",
       "      <td>15870</td>\n",
       "      <td>2.1522684310018896</td>\n",
       "      <td>1.7581135042599598</td>\n",
       "      <td>0.0</td>\n",
       "      <td>19.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rain_accumulation</th>\n",
       "      <td>15872</td>\n",
       "      <td>2.4571572580650005E-4</td>\n",
       "      <td>0.0057790024267639</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.43</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rain_duration</th>\n",
       "      <td>15872</td>\n",
       "      <td>0.3811743951612903</td>\n",
       "      <td>4.38803609221701</td>\n",
       "      <td>0.0</td>\n",
       "      <td>60.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>relative_humidity</th>\n",
       "      <td>15873</td>\n",
       "      <td>47.599313299313295</td>\n",
       "      <td>26.194044531313825</td>\n",
       "      <td>1.5</td>\n",
       "      <td>92.9</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                        0                      1                   2      3  \\\n",
       "summary             count                   mean              stddev    min   \n",
       "rowID               15873               793600.0   458228.4746717515      0   \n",
       "air_pressure        15873      916.8291627291593  3.0517222151797627  905.1   \n",
       "air_temp            15873      61.85468909468877  11.835413790821452  32.36   \n",
       "avg_wind_direction  15870      161.2875236294896   95.31316129656483    0.0   \n",
       "avg_wind_speed      15870      2.792804032766223    2.07050619846002    0.1   \n",
       "max_wind_direction  15870     162.70094517958412   92.26960112663168    0.0   \n",
       "max_wind_speed      15870     3.4146250787649493  2.4289064068121387    0.1   \n",
       "min_wind_direction  15870     166.64429741650915   97.82483630682518    0.0   \n",
       "min_wind_speed      15870     2.1522684310018896  1.7581135042599598    0.0   \n",
       "rain_accumulation   15872  2.4571572580650005E-4  0.0057790024267639    0.0   \n",
       "rain_duration       15872     0.3811743951612903    4.38803609221701    0.0   \n",
       "relative_humidity   15873     47.599313299313295  26.194044531313825    1.5   \n",
       "\n",
       "                          4  \n",
       "summary                 max  \n",
       "rowID               1587200  \n",
       "air_pressure          929.4  \n",
       "air_temp              96.44  \n",
       "avg_wind_direction    359.0  \n",
       "avg_wind_speed         20.1  \n",
       "max_wind_direction    359.0  \n",
       "max_wind_speed         20.9  \n",
       "min_wind_direction    359.0  \n",
       "min_wind_speed         19.5  \n",
       "rain_accumulation      0.43  \n",
       "rain_duration          60.0  \n",
       "relative_humidity      92.9  "
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Show summary statistics\n",
    "\n",
    "filteredDF.describe().toPandas().transpose()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "15869"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Drop null data\n",
    "\n",
    "workingDF = filteredDF.na.drop()\n",
    "workingDF.count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true,
    "jupyter": {
     "outputs_hidden": true
    }
   },
   "outputs": [],
   "source": [
    "# Create feature vector\n",
    "\n",
    "featuresUsed = ['air_pressure', 'air_temp', 'avg_wind_direction', 'avg_wind_speed', 'max_wind_direction', \n",
    "        'max_wind_speed','relative_humidity']\n",
    "assembler = VectorAssembler(inputCols=featuresUsed, outputCol=\"features_unscaled\")\n",
    "assembled = assembler.transform(workingDF)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true,
    "jupyter": {
     "outputs_hidden": true
    }
   },
   "outputs": [],
   "source": [
    "# Scale data\n",
    "\n",
    "scaler = StandardScaler(inputCol=\"features_unscaled\", outputCol=\"features\", withStd=True, withMean=True)\n",
    "scalerModel = scaler.fit(assembled)\n",
    "scaledData = scalerModel.transform(assembled)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "5289"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Use one-third data for elbow plot\n",
    "\n",
    "scaledData = scaledData.select(\"features\", \"rowID\")\n",
    "\n",
    "elbowset = scaledData.filter((scaledData.rowID % 3) == 0).select(\"features\")\n",
    "elbowset.persist()\n",
    "elbowset.count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training for cluster size 5 \n",
      "......................WSSE = 8804.694339031907 \n",
      "Training for cluster size 6 \n",
      "......................WSSE = 8525.480223872479 \n",
      "Training for cluster size 7 \n",
      "......................WSSE = 8250.724863437203 \n",
      "Training for cluster size 8 \n",
      "......................WSSE = 7848.19734934203 \n",
      "Training for cluster size 9 \n",
      "......................WSSE = 7570.8780978830055 \n",
      "Training for cluster size 10 \n",
      "......................WSSE = 7329.067961103764 \n",
      "Training for cluster size 11 \n",
      "......................WSSE = 7177.965311120956 \n",
      "Training for cluster size 12 \n",
      "......................WSSE = 7065.723853555635 \n",
      "Training for cluster size 13 \n",
      "......................WSSE = 7082.4094400840595 \n",
      "Training for cluster size 14 \n",
      "......................WSSE = 6802.849359722116 \n",
      "Training for cluster size 15 \n",
      "......................WSSE = 6679.158807094141 \n",
      "Training for cluster size 16 \n",
      "......................WSSE = 6748.620942357968 \n",
      "Training for cluster size 17 \n",
      "......................WSSE = 6528.072121516723 \n",
      "Training for cluster size 18 \n",
      "......................WSSE = 6400.743795027434 \n",
      "Training for cluster size 19 \n",
      "......................WSSE = 6332.101590011742 \n"
     ]
    }
   ],
   "source": [
    "# Generate clusters for elbow plot\n",
    "\n",
    "clusters = range(5,20)\n",
    "wsseList = utils.elbow(elbowset, clusters)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Show elbow plot\n",
    "\n",
    "utils.elbow_plot(wsseList, clusters)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "jupyter": {
     "outputs_hidden": true
    }
   },
   "outputs": [],
   "source": [
    "# Run KMeans for k = 12\n",
    "\n",
    "scaledDataFeat = scaledData.select(\"features\")\n",
    "scaledDataFeat.persist()\n",
    "\n",
    "kmeans = KMeans(k=12, seed=1)\n",
    "model = kmeans.fit(scaledDataFeat)\n",
    "transformed = model.transform(scaledDataFeat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    },
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Compute cluster centers\n",
    "\n",
    "centers = model.clusterCenters()\n",
    "P = utils.pd_centers(featuresUsed, centers)\n",
    "centers"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dry Days"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "utils.parallel_plot(P[P['relative_humidity'] < -0.5], P)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Humid Days"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "utils.parallel_plot(P[P['relative_humidity'] > 0.5], P)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Hot Days"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "utils.parallel_plot(P[P['air_temp'] > 0.5], P)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Cool Days"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "utils.parallel_plot(P[P['air_temp'] < -0.5], P)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "jupyter": {
     "outputs_hidden": true
    }
   },
   "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": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}