"
+ ],
+ "text/plain": [
+ " zn indus chas rm age dis tax ptratio black lstat medv\n",
+ "0 0.0 10.81 0.0 5.961 17.5 5.2873 305.0 19.2 376.94 9.88 21.7\n",
+ "1 25.0 5.13 0.0 5.927 47.2 6.9320 284.0 19.7 396.90 9.22 19.6\n",
+ "2 0.0 9.90 0.0 5.972 76.7 3.1025 304.0 18.4 396.24 9.97 20.3\n",
+ "3 0.0 19.58 0.0 5.597 94.9 1.5257 403.0 14.7 351.85 21.45 15.4\n",
+ "4 21.0 5.64 0.0 6.115 63.0 6.8147 243.0 16.8 393.97 9.43 20.5"
+ ]
+ },
+ "execution_count": 68,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
+ "source": [
+ "boston_nocorr.head()"
]
},
{
@@ -100,11 +819,30 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 69,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ "count 398.000000\n",
+ "mean 22.459799\n",
+ "std 8.774737\n",
+ "min 5.000000\n",
+ "25% 17.325000\n",
+ "50% 21.550000\n",
+ "75% 25.000000\n",
+ "max 50.000000\n",
+ "Name: medv, dtype: float64"
+ ]
+ },
+ "execution_count": 69,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
"source": [
- "# Your code here"
+ "boston_nocorr[\"medv\"].describe()"
]
},
{
@@ -126,16 +864,24 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 70,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "from sklearn.metrics import r2_score"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 72,
"metadata": {},
"outputs": [],
"source": [
- "from sklearn.metrics import r2_score\n",
- "\n",
"def performance_metric(y_true, y_predict):\n",
" \"\"\" Calculates and returns the performance score between \n",
" true and predicted values based on the metric chosen. \"\"\"\n",
- " # Your code here:"
+ " score = r2_score(y_true, y_predict)\n",
+ " return score"
]
},
{
@@ -148,11 +894,22 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 73,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "X = boston_nocorr.drop(columns = \"medv\")\n",
+ "y = boston_nocorr[\"medv\"]"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 75,
"metadata": {},
"outputs": [],
"source": [
- "# Your code here"
+ "from sklearn.model_selection import train_test_split\n",
+ "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=8)"
]
},
{
@@ -175,11 +932,151 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 79,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "from sklearn.ensemble import RandomForestRegressor"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 101,
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stderr",
+ "output_type": "stream",
+ "text": [
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n"
+ ]
+ },
+ {
+ "name": "stderr",
+ "output_type": "stream",
+ "text": [
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n"
+ ]
+ }
+ ],
+ "source": [
+ "depths = range(1,50)\n",
+ "train = []\n",
+ "test = []\n",
+ "for i in depths:\n",
+ " RF = RandomForestRegressor(max_depth = i)\n",
+ " RF.fit(X_train, y_train)\n",
+ " y_train_pred = RF.predict(X_train)\n",
+ " y_test_pred = RF.predict(X_test)\n",
+ " train.append(performance_metric(y_train, y_train_pred))\n",
+ " test.append(performance_metric(y_test, y_test_pred))\n"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 102,
"metadata": {},
"outputs": [],
"source": [
- "# Five separate RFR here with the given max depths"
+ "df = pd.DataFrame({\"depth\":depths,\n",
+ " \"train\":train,\n",
+ " \"test\":test})"
]
},
{
@@ -191,13 +1088,68 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 103,
"metadata": {
- "scrolled": false
+ "scrolled": true
},
- "outputs": [],
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ ""
+ ]
+ },
+ "execution_count": 103,
+ "metadata": {},
+ "output_type": "execute_result"
+ },
+ {
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ ""
+ ]
+ },
+ "metadata": {
+ "needs_background": "light"
+ },
+ "output_type": "display_data"
+ }
+ ],
+ "source": [
+ "df.plot.scatter(x='depth', y='train')"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 104,
+ "metadata": {},
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ ""
+ ]
+ },
+ "execution_count": 104,
+ "metadata": {},
+ "output_type": "execute_result"
+ },
+ {
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ ""
+ ]
+ },
+ "metadata": {
+ "needs_background": "light"
+ },
+ "output_type": "display_data"
+ }
+ ],
"source": [
- "# Produce a plot with the score for the testing and training for the different max depths"
+ "df.plot.scatter(x='depth', y='test')"
]
},
{
@@ -209,11 +1161,11 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 92,
"metadata": {},
"outputs": [],
"source": [
- "# Your response here"
+ "#There is a moment while increasing depth where even though in the train set the score increases, the score in the test set decreases."
]
},
{
@@ -226,11 +1178,11 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 105,
"metadata": {},
"outputs": [],
"source": [
- "# Your response here"
+ "# When maximum depth is 1, the model is too simplistic and it is suffering from high bias, while if the maximum depth is 10 we are in risk of high variance."
]
},
{
@@ -243,11 +1195,82 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 107,
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stderr",
+ "output_type": "stream",
+ "text": [
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
+ "/opt/anaconda3/lib/python3.7/site-packages/sklearn/ensemble/forest.py:245: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
+ " \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n"
+ ]
+ },
+ {
+ "data": {
+ "text/plain": [
+ ""
+ ]
+ },
+ "execution_count": 107,
+ "metadata": {},
+ "output_type": "execute_result"
+ },
+ {
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ ""
+ ]
+ },
+ "metadata": {
+ "needs_background": "light"
+ },
+ "output_type": "display_data"
+ }
+ ],
+ "source": [
+ "depths = range(1,11)\n",
+ "train = []\n",
+ "test = []\n",
+ "for i in depths:\n",
+ " RF = RandomForestRegressor(max_depth = i)\n",
+ " RF.fit(X_train, y_train)\n",
+ " y_train_pred = RF.predict(X_train)\n",
+ " y_test_pred = RF.predict(X_test)\n",
+ " train.append(performance_metric(y_train, y_train_pred))\n",
+ " test.append(performance_metric(y_test, y_test_pred))\n",
+ "df = pd.DataFrame({\"depth\":depths,\n",
+ " \"train\":train,\n",
+ " \"test\":test})\n",
+ "df.plot.scatter(x='depth', y='test')"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 108,
"metadata": {},
"outputs": [],
"source": [
- "# Your response here"
+ "#I would use 5 as maximum depth, because I think is the inflection point where the model stops improving over the test set and the high variance risk increases."
]
},
{
@@ -265,12 +1288,22 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 109,
"metadata": {},
"outputs": [],
"source": [
- "# Your response here"
+ "# We always need to take into account the origin of the data. If the data is from 1978 and it is about house price we cannot rely on the model for actual predictions.\n",
+ "# There are many features that are determining the price house, and in this datset we have just a few.\n",
+ "# I don't think it is robus enough\n",
+ "# Urban cities and rural cities having different conditions and features, so we can not use this model to rural cities predictions"
]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": []
}
],
"metadata": {
@@ -290,7 +1323,20 @@
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
- "version": "3.7.2"
+ "version": "3.7.4"
+ },
+ "toc": {
+ "base_numbering": 1,
+ "nav_menu": {},
+ "number_sections": true,
+ "sideBar": true,
+ "skip_h1_title": false,
+ "title_cell": "Table of Contents",
+ "title_sidebar": "Contents",
+ "toc_cell": false,
+ "toc_position": {},
+ "toc_section_display": true,
+ "toc_window_display": false
}
},
"nbformat": 4,
diff --git a/your-code/lab_overfitting.ipynb b/your-code/lab_overfitting.ipynb
index e0eafec..1961897 100644
--- a/your-code/lab_overfitting.ipynb
+++ b/your-code/lab_overfitting.ipynb
@@ -21,7 +21,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
@@ -55,11 +55,20 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 2,
"metadata": {},
"outputs": [],
"source": [
- "# Your code here\n"
+ "from sklearn.svm import SVC"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 8,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "classifiers = [SVC(gamma = 0.001), SVC(gamma = 1), SVC(gamma = 20)]"
]
},
{
@@ -71,9 +80,20 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 9,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ ""
+ ]
+ },
+ "metadata": {},
+ "output_type": "display_data"
+ }
+ ],
"source": [
"from matplotlib.colors import ListedColormap\n",
"\n",
@@ -85,7 +105,8 @@
"xx, yy = np.meshgrid(np.arange(x_min, x_max, h),\n",
" np.arange(y_min, y_max, h))\n",
"cm = plt.cm.RdBu\n",
- "cm_bright = ListedColormap(['#FF0000', '#0000FF'])\nnames = ['gamma = 0.001', 'gamma = 1', 'gamma = 20']\n",
+ "cm_bright = ListedColormap(['#FF0000', '#0000FF'])\n",
+ "names = ['gamma = 0.001', 'gamma = 1', 'gamma = 20']\n",
"\n",
"# iterate over classifiers\n",
"for name, clf in zip(names, classifiers):\n",
@@ -132,7 +153,7 @@
"metadata": {},
"outputs": [],
"source": [
- "# Your response here"
+ "# I think gamma 20 fits the data best, but the model is too complex, I would select gamma 1"
]
},
{
@@ -144,11 +165,24 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 12,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ "0.93"
+ ]
+ },
+ "execution_count": 12,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
"source": [
- "# Your code here"
+ "model = SVC(gamma = 0.7)\n",
+ "model.fit(X, y)\n",
+ "model.score(X, y)"
]
},
{
@@ -160,11 +194,56 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 53,
"metadata": {},
"outputs": [],
"source": [
- "# Your code here"
+ "from sklearn.model_selection import train_test_split\n",
+ "X_train, X_test, y_train, y_test = train_test_split(X, y,\n",
+ " test_size=0.2,\n",
+ " random_state=500)"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 54,
+ "metadata": {},
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ "0.975"
+ ]
+ },
+ "execution_count": 54,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
+ "source": [
+ "model = SVC(gamma = 20)\n",
+ "model.fit(X_train, y_train)\n",
+ "model.score(X_train, y_train)"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 55,
+ "metadata": {},
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ "0.8"
+ ]
+ },
+ "execution_count": 55,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
+ "source": [
+ "model.score(X_test, y_test)"
]
},
{
@@ -176,11 +255,23 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 70,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "0.9375\n",
+ "0.8\n"
+ ]
+ }
+ ],
"source": [
- "# Your code here"
+ "model = SVC(gamma = 4)\n",
+ "model.fit(X_train, y_train)\n",
+ "print(model.score(X_train, y_train))\n",
+ "print(model.score(X_test, y_test))"
]
},
{
@@ -196,7 +287,7 @@
"metadata": {},
"outputs": [],
"source": [
- "# Your response here"
+ "# It was, but not too much, in fact reducing the gamma is not improving the score over the test set."
]
}
],
@@ -216,7 +307,20 @@
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
- "version": "3.7.2"
+ "version": "3.7.4"
+ },
+ "toc": {
+ "base_numbering": 1,
+ "nav_menu": {},
+ "number_sections": true,
+ "sideBar": true,
+ "skip_h1_title": false,
+ "title_cell": "Table of Contents",
+ "title_sidebar": "Contents",
+ "toc_cell": false,
+ "toc_position": {},
+ "toc_section_display": true,
+ "toc_window_display": false
}
},
"nbformat": 4,