Removing lock

Schrick-Noah_Learning-Practice-8.ipynb

@@ -14,7 +14,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 16,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -27,12 +27,15 @@
     "import pandas as pd\n",
     "from sklearn.model_selection import train_test_split\n",
     "from sklearn.naive_bayes import MultinomialNB\n",
+    "from sklearn.metrics import mean_squared_error\n",
+    "from math import sqrt\n",
+    "\n",
     "import matplotlib.pylab as plt\n",
     "from dmba import classificationSummary, gainsChart\n",
     "\n",
     "from sklearn import preprocessing\n",
     "from sklearn.metrics import accuracy_score\n",
-    "from sklearn.neighbors import NearestNeighbors, KNeighborsClassifier"
+    "from sklearn.neighbors import NearestNeighbors, KNeighborsClassifier, KNeighborsRegressor"
    ]
   },
   {
@@ -49,11 +52,13 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 64,
    "metadata": {},
    "outputs": [],
    "source": [
-    "# Pre-processing"
+    "# Pre-processing\n",
+    "housing_df = pd.read_csv('BostonHousing.csv')\n",
+    "trainData, validData = train_test_split(housing_df, test_size=0.4, random_state=26)"
    ]
   },
   {
@@ -67,11 +72,72 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 68,
    "metadata": {},
    "outputs": [],
    "source": [
-    "# a"
+    "# a\n",
+    "## Normalize\n",
+    "scaler = preprocessing.StandardScaler()\n",
+    "\n",
+    "predictors = housing_df.columns.values.tolist()\n",
+    "predictors.remove('CAT. MEDV')\n",
+    "predictors.remove('MEDV')\n",
+    "\n",
+    "scaler.fit(trainData[predictors])  # Note the use of an array of column names\n",
+    "\n",
+    "# Transform the full dataset\n",
+    "housingNorm = pd.concat([pd.DataFrame(scaler.transform(housing_df[predictors]), \n",
+    "                                    columns=predictors),\n",
+    "                        housing_df[['MEDV']]], axis=1)\n",
+    "\n",
+    "trainNorm = housingNorm.iloc[trainData.index]\n",
+    "validNorm = housingNorm.iloc[validData.index]\n",
+    "#newHousingNorm = pd.DataFrame(scaler.transform(housingNorm), columns=predictors)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 69,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "   k  accuracy\n",
+      "0  1  0.643141\n",
+      "1  2  0.720241\n",
+      "2  3  0.710570\n",
+      "3  4  0.676896\n"
+     ]
+    }
+   ],
+   "source": [
+    "## K-NN\n",
+    "results = []\n",
+    "train_X = trainNorm[predictors]\n",
+    "train_y = trainNorm['MEDV']\n",
+    "valid_X = validNorm[predictors]\n",
+    "valid_y = validNorm['MEDV']\n",
+    "\n",
+    "for k in range(1,5):\n",
+    "    knn = KNeighborsRegressor(n_neighbors=k).fit(train_X, train_y)\n",
+    "    results.append({\n",
+    "        'k': k,\n",
+    "        'accuracy': knn.score(valid_X, valid_y)\n",
+    "    })\n",
+    "\n",
+    "# Convert results to a pandas data frame\n",
+    "results = pd.DataFrame(results)\n",
+    "print(results)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "The best k is k=2. This means we have greater accuracy when each point has its nearest 2 neighbors identified."
    ]
   },
   {
@@ -87,11 +153,33 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 73,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "[19.45]\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/home/noah/.local/lib/python3.10/site-packages/sklearn/base.py:420: UserWarning: X does not have valid feature names, but StandardScaler was fitted with feature names\n",
+      "  warnings.warn(\n",
+      "/home/noah/.local/lib/python3.10/site-packages/sklearn/base.py:420: UserWarning: X does not have valid feature names, but KNeighborsRegressor was fitted with feature names\n",
+      "  warnings.warn(\n"
+     ]
+    }
+   ],
    "source": [
-    "# b"
+    "# b\n",
+    "sample = [[0.2, 0, 7, 0, 0.538, 6, 62, 4.7, 4, 307, 21, 10]]\n",
+    "sample_norm = scaler.transform(sample)\n",
+    "ans_b = knn.predict(sample_norm)\n",
+    "print(ans_b)"
    ]
   },
   {
@@ -104,12 +192,11 @@
    ]
   },
   {
-   "cell_type": "code",
-   "execution_count": null,
+   "attachments": {},
+   "cell_type": "markdown",
    "metadata": {},
-   "outputs": [],
    "source": [
-    "# c"
+    "Error will be zero, or close to zero. Since the training data was used to build the model, checking the training data against itself should yield very high accuracy."
    ]
   },
   {
@@ -126,7 +213,8 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# d."
+    "# d.\n",
+    "This model was built and tuned with a specific k value (k=2). This value of k was the best for the data used, but does not mean that this value of k is the best for future data."
    ]
   },
   {
@@ -143,7 +231,14 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# e."
+    "# e.\n",
+    "\n",
+    "1. Select k-nearest samples\n",
+    "2. Compute average value for k-nearest neighbors\n",
+    "3. Score value with new data\n",
+    "4. Repeat for all new data samples\n",
+    "\n",
+    "If there are several thousands of new tracts, then k-NN will take much longer to run. The model will need to be rebuilt numerous times over the several thousand pieces of data. The lack of scalability will be a disadvantage. "
    ]
   },
   {
@@ -388,8 +483,22 @@
   }
  ],
  "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
   "language_info": {
-   "name": "python"
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.10.10"
   },
   "orig_nbformat": 4
  },