minor polish on pipeline stuff, use bunch instead of dictionary on na…

…med_steps
amueller · May 3, 2018 · c417bda · c417bda
1 parent 0259ecc
commit c417bda
Showing 1 changed file with 10 additions and 19 deletions.
diff --git a/notebooks/02 Pipelines.ipynb b/notebooks/02 Pipelines.ipynb
@@ -244,7 +244,7 @@
  " # iterate over all but the final step\n",
  " # transform the data\n",
  " X_transformed = step[1].transform(X_transformed)\n",
- " # fit the last step\n",
+ " # predict using the last step\n",
  " return self.steps[-1][1].predict(X_transformed)"
  ]
  },
@@ -313,7 +313,7 @@
  "# fit the pipeline defined above to the cancer dataset\n",
  "pipe.fit(cancer.data)\n",
  "# extract the first two principal components from the \"pca\" step\n",
- "components = pipe.named_steps[\"pca\"].components_\n",
+ "components = pipe.named_steps.pca.components_\n",
  "print(components.shape)"
  ]
  },
@@ -380,16 +380,7 @@
  "metadata": {},
  "outputs": [],
  "source": [
- "print(grid.best_estimator_.named_steps[\"logisticregression\"])"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": null,
- "metadata": {},
- "outputs": [],
- "source": [
- "print(grid.best_estimator_.named_steps[\"logisticregression\"].coef_)"
+ "print(grid.best_estimator_.named_steps.logisticregression)"
  ]
  },
  {
@@ -398,7 +389,7 @@
  "metadata": {},
  "outputs": [],
  "source": [
- "grid.best_estimator_.named_steps.logisticregression.coef_"
+ "print(grid.best_estimator_.named_steps.logisticregression.coef_)"
  ]
  },
  {
@@ -524,9 +515,7 @@
  "metadata": {},
  "outputs": [],
  "source": [
- "param_grid = [{'regressor': [DecisionTreeRegressor()], 'regressor__max_depth': [2, 3, 4]},\n",
- " {'regressor': [Ridge()], 'regressor__alpha': [0.1, 1]}\n",
- " ]"
+ "grid.best_params_"
  ]
  },
  {
@@ -535,7 +524,9 @@
  "metadata": {},
  "outputs": [],
  "source": [
- "grid.best_params_"
+ "param_grid = [{'regressor': [DecisionTreeRegressor()], 'regressor__max_depth': [2, 3, 4]},\n",
+ " {'regressor': [Ridge()], 'regressor__alpha': [0.1, 1]}\n",
+ " ]"
  ]
  },
  {
@@ -544,7 +535,7 @@
  "source": [
  "# Exercise\n",
  "\n",
- "Load the bike dataset from notebook 1. Create a pipline using one-hot-encoding, scaling,  and polynomial features and a linear regression model (like ridge or lasso).\n",
+ "Load the adult. Create a pipline using scaling, and polynomial features and a linear classifier.\n",
  "\n",
  "Search over the best options for the polynomial features together with the regularization of a linear model."
  ]
@@ -574,7 +565,7 @@
  "name": "python",
  "nbconvert_exporter": "python",
  "pygments_lexer": "ipython3",
- "version": "3.6.1"
+ "version": "3.6.3"
  }
  },
  "nbformat": 4,