cmap=plt.cm.bone,edgecolor='black')

clf=linear_model.SGDClassifier(alpha=0.01,max_iter=100,tol=1e-3)

clf=linear_model.SGDClassifier(alpha=0.01,max_iter=100)

clf.fit(X,y)

Z=clf.decision_function(np.c_[xx.ravel(),yy.ravel()])

Z=Z.reshape(xx.shape)

no_weights=plt.contour(xx,yy,Z,levels=[0],linestyles=['solid'])

clf=linear_model.SGDClassifier(alpha=0.01,max_iter=100,tol=1e-3)

clf=linear_model.SGDClassifier(alpha=0.01,max_iter=100)

clf.fit(X,y,sample_weight=sample_weight)

Z=clf.decision_function(np.c_[xx.ravel(),yy.ravel()])

Z=Z.reshape(xx.shape)

},

"outputs": [],

"source": [

"# Author: Rob Zinkov <rob at zinkov dot com>\n# License: BSD 3 clause\n\nimport numpy as np\nimport matplotlib.pyplot as plt\nfrom sklearn import datasets\n\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.linear_model import SGDClassifier, Perceptron\nfrom sklearn.linear_model import PassiveAggressiveClassifier\nfrom sklearn.linear_model import LogisticRegression\n\nheldout = [0.95, 0.90, 0.75, 0.50, 0.01]\nrounds = 20\nX, y = datasets.load_digits(return_X_y=True)\n\nclassifiers = [\n (\"SGD\", SGDClassifier(max_iter=100, tol=1e-3)),\n (\"ASGD\", SGDClassifier(average=True, max_iter=1000, tol=1e-3)),\n (\"Perceptron\", Perceptron(tol=1e-3)),\n (\"Passive-Aggressive I\", PassiveAggressiveClassifier(loss='hinge',\n C=1.0, tol=1e-4)),\n (\"Passive-Aggressive II\", PassiveAggressiveClassifier(loss='squared_hinge',\n C=1.0, tol=1e-4)),\n (\"SAG\", LogisticRegression(solver='sag', tol=1e-1, C=1.e4 / X.shape[0]))\n]\n\nxx = 1. - np.array(heldout)\n\nfor name, clf in classifiers:\n print(\"training %s\" % name)\n rng = np.random.RandomState(42)\n yy = []\n for i in heldout:\n yy_ = []\n for r in range(rounds):\n X_train, X_test, y_train, y_test = \\\n train_test_split(X, y, test_size=i, random_state=rng)\n clf.fit(X_train, y_train)\n y_pred = clf.predict(X_test)\n yy_.append(1 - np.mean(y_pred == y_test))\n yy.append(np.mean(yy_))\n plt.plot(xx, yy, label=name)\n\nplt.legend(loc=\"upper right\")\nplt.xlabel(\"Proportion train\")\nplt.ylabel(\"Test Error Rate\")\nplt.show()"

"# Author: Rob Zinkov <rob at zinkov dot com>\n# License: BSD 3 clause\n\nimport numpy as np\nimport matplotlib.pyplot as plt\nfrom sklearn import datasets\n\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.linear_model import SGDClassifier, Perceptron\nfrom sklearn.linear_model import PassiveAggressiveClassifier\nfrom sklearn.linear_model import LogisticRegression\n\nheldout = [0.95, 0.90, 0.75, 0.50, 0.01]\nrounds = 20\nX, y = datasets.load_digits(return_X_y=True)\n\nclassifiers = [\n (\"SGD\", SGDClassifier(max_iter=100)),\n (\"ASGD\", SGDClassifier(average=True, max_iter=1000)),\n (\"Perceptron\", Perceptron(tol=1e-3)),\n (\"Passive-Aggressive I\", PassiveAggressiveClassifier(loss='hinge',\n C=1.0, tol=1e-4)),\n (\"Passive-Aggressive II\", PassiveAggressiveClassifier(loss='squared_hinge',\n C=1.0, tol=1e-4)),\n (\"SAG\", LogisticRegression(solver='sag', tol=1e-1, C=1.e4 / X.shape[0]))\n]\n\nxx = 1. - np.array(heldout)\n\nfor name, clf in classifiers:\n print(\"training %s\" % name)\n rng = np.random.RandomState(42)\n yy = []\n for i in heldout:\n yy_ = []\n for r in range(rounds):\n X_train, X_test, y_train, y_test = \\\n train_test_split(X, y, test_size=i, random_state=rng)\n clf.fit(X_train, y_train)\n y_pred = clf.predict(X_test)\n yy_.append(1 - np.mean(y_pred == y_test))\n yy.append(np.mean(yy_))\n plt.plot(xx, yy, label=name)\n\nplt.legend(loc=\"upper right\")\nplt.xlabel(\"Proportion train\")\nplt.ylabel(\"Test Error Rate\")\nplt.show()"

]

}

],

clf=SGDClassifier(loss="hinge",alpha=0.01,max_iter=200,

fit_intercept=True,tol=1e-3)

fit_intercept=True)

clf.fit(X,Y)

X,y=datasets.load_digits(return_X_y=True)

classifiers= [

("SGD",SGDClassifier(max_iter=100,tol=1e-3)),

("ASGD",SGDClassifier(average=True,max_iter=1000,tol=1e-3)),

("SGD",SGDClassifier(max_iter=100)),

("ASGD",SGDClassifier(average=True,max_iter=1000)),

("Perceptron",Perceptron(tol=1e-3)),

("Passive-Aggressive I",PassiveAggressiveClassifier(loss='hinge',

C=1.0,tol=1e-4)),

pipeline=Pipeline([

('vect',CountVectorizer()),

('tfidf',TfidfTransformer()),

('clf',SGDClassifier(tol=1e-3)),

('clf',SGDClassifier()),

])

clf=SGDClassifier(alpha=0.001,max_iter=100,tol=1e-3).fit(X,y)

clf=SGDClassifier(alpha=0.001,max_iter=100).fit(X,y)

x_min,x_max=X[:,0].min()-1,X[:,0].max()+1

},

"outputs": [],

"source": [

"print(__doc__)\n\nimport numpy as np\nimport matplotlib.pyplot as plt\nfrom sklearn import linear_model\n\n# we create 20 points\nnp.random.seed(0)\nX = np.r_[np.random.randn(10, 2) + [1, 1], np.random.randn(10, 2)]\ny = [1] * 10 + [-1] * 10\nsample_weight = 100 * np.abs(np.random.randn(20))\n# and assign a bigger weight to the last 10 samples\nsample_weight[:10] *= 10\n\n# plot the weighted data points\nxx, yy = np.meshgrid(np.linspace(-4, 5, 500), np.linspace(-4, 5, 500))\nplt.figure()\nplt.scatter(X[:, 0], X[:, 1], c=y, s=sample_weight, alpha=0.9,\n cmap=plt.cm.bone, edgecolor='black')\n\n# fit the unweighted model\nclf = linear_model.SGDClassifier(alpha=0.01, max_iter=100, tol=1e-3)\nclf.fit(X, y)\nZ = clf.decision_function(np.c_[xx.ravel(), yy.ravel()])\nZ = Z.reshape(xx.shape)\nno_weights = plt.contour(xx, yy, Z, levels=[0], linestyles=['solid'])\n\n# fit the weighted model\nclf = linear_model.SGDClassifier(alpha=0.01, max_iter=100, tol=1e-3)\nclf.fit(X, y, sample_weight=sample_weight)\nZ = clf.decision_function(np.c_[xx.ravel(), yy.ravel()])\nZ = Z.reshape(xx.shape)\nsamples_weights = plt.contour(xx, yy, Z, levels=[0], linestyles=['dashed'])\n\nplt.legend([no_weights.collections[0], samples_weights.collections[0]],\n [\"no weights\", \"with weights\"], loc=\"lower left\")\n\nplt.xticks(())\nplt.yticks(())\nplt.show()"

"print(__doc__)\n\nimport numpy as np\nimport matplotlib.pyplot as plt\nfrom sklearn import linear_model\n\n# we create 20 points\nnp.random.seed(0)\nX = np.r_[np.random.randn(10, 2) + [1, 1], np.random.randn(10, 2)]\ny = [1] * 10 + [-1] * 10\nsample_weight = 100 * np.abs(np.random.randn(20))\n# and assign a bigger weight to the last 10 samples\nsample_weight[:10] *= 10\n\n# plot the weighted data points\nxx, yy = np.meshgrid(np.linspace(-4, 5, 500), np.linspace(-4, 5, 500))\nplt.figure()\nplt.scatter(X[:, 0], X[:, 1], c=y, s=sample_weight, alpha=0.9,\n cmap=plt.cm.bone, edgecolor='black')\n\n# fit the unweighted model\nclf = linear_model.SGDClassifier(alpha=0.01, max_iter=100)\nclf.fit(X, y)\nZ = clf.decision_function(np.c_[xx.ravel(), yy.ravel()])\nZ = Z.reshape(xx.shape)\nno_weights = plt.contour(xx, yy, Z, levels=[0], linestyles=['solid'])\n\n# fit the weighted model\nclf = linear_model.SGDClassifier(alpha=0.01, max_iter=100)\nclf.fit(X, y, sample_weight=sample_weight)\nZ = clf.decision_function(np.c_[xx.ravel(), yy.ravel()])\nZ = Z.reshape(xx.shape)\nsamples_weights = plt.contour(xx, yy, Z, levels=[0], linestyles=['dashed'])\n\nplt.legend([no_weights.collections[0], samples_weights.collections[0]],\n [\"no weights\", \"with weights\"], loc=\"lower left\")\n\nplt.xticks(())\nplt.yticks(())\nplt.show()"

]

}

],

print("Computing principal eigenvector score using a power iteration method")

t0=time()

scores=centrality_scores(X,max_iter=100,tol=1e-10)

scores=centrality_scores(X,max_iter=100)

print("done in %0.3fs"% (time()-t0))

pprint([names[i]foriinnp.abs(scores).argsort()[-10:]])

partial_fit_classifiers= {

'SGD':SGDClassifier(max_iter=5,tol=1e-3),

'SGD':SGDClassifier(max_iter=5),

'Perceptron':Perceptron(tol=1e-3),

'NB Multinomial':MultinomialNB(alpha=0.01),

'Passive-Aggressive':PassiveAggressiveClassifier(tol=1e-3),

estimator_dict= {

'No stopping criterion':

linear_model.SGDClassifier(tol=1e-3,n_iter_no_change=3),

linear_model.SGDClassifier(n_iter_no_change=3),

'Training loss':

linear_model.SGDClassifier(early_stopping=False,n_iter_no_change=3,

tol=0.1),

},

"outputs": [],

"source": [

"# Authors: Tom Dupre la Tour\n#\n# License: BSD 3 clause\nimport time\nimport sys\n\nimport pandas as pd\nimport numpy as np\nimport matplotlib.pyplot as plt\n\nfrom sklearn import linear_model\nfrom sklearn.datasets import fetch_openml\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.utils.testing import ignore_warnings\nfrom sklearn.exceptions import ConvergenceWarning\nfrom sklearn.utils import shuffle\n\nprint(__doc__)\n\n\ndef load_mnist(n_samples=None, class_0='0', class_1='8'):\n \"\"\"Load MNIST, select two classes, shuffle and return only n_samples.\"\"\"\n # Load data from http://openml.org/d/554\n mnist = fetch_openml('mnist_784', version=1)\n\n # take only two classes for binary classification\n mask = np.logical_or(mnist.target == class_0, mnist.target == class_1)\n\n X, y = shuffle(mnist.data[mask], mnist.target[mask], random_state=42)\n if n_samples is not None:\n X, y = X[:n_samples], y[:n_samples]\n return X, y\n\n\n@ignore_warnings(category=ConvergenceWarning)\ndef fit_and_score(estimator, max_iter, X_train, X_test, y_train, y_test):\n \"\"\"Fit the estimator on the train set and score it on both sets\"\"\"\n estimator.set_params(max_iter=max_iter)\n estimator.set_params(random_state=0)\n\n start = time.time()\n estimator.fit(X_train, y_train)\n\n fit_time = time.time() - start\n n_iter = estimator.n_iter_\n train_score = estimator.score(X_train, y_train)\n test_score = estimator.score(X_test, y_test)\n\n return fit_time, n_iter, train_score, test_score\n\n\n# Define the estimators to compare\nestimator_dict = {\n 'No stopping criterion':\n linear_model.SGDClassifier(tol=1e-3, n_iter_no_change=3),\n 'Training loss':\n linear_model.SGDClassifier(early_stopping=False, n_iter_no_change=3,\n tol=0.1),\n 'Validation score':\n linear_model.SGDClassifier(early_stopping=True, n_iter_no_change=3,\n tol=0.0001, validation_fraction=0.2)\n}\n\n# Load the dataset\nX, y = load_mnist(n_samples=10000)\nX_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.5,\n random_state=0)\n\nresults = []\nfor estimator_name, estimator in estimator_dict.items():\n print(estimator_name + ': ', end='')\n for max_iter in range(1, 50):\n print('.', end='')\n sys.stdout.flush()\n\n fit_time, n_iter, train_score, test_score = fit_and_score(\n estimator, max_iter, X_train, X_test, y_train, y_test)\n\n results.append((estimator_name, max_iter, fit_time, n_iter,\n train_score, test_score))\n print('')\n\n# Transform the results in a pandas dataframe for easy plotting\ncolumns = [\n 'Stopping criterion', 'max_iter', 'Fit time (sec)', 'n_iter_',\n 'Train score', 'Test score'\n]\nresults_df = pd.DataFrame(results, columns=columns)\n\n# Define what to plot (x_axis, y_axis)\nlines = 'Stopping criterion'\nplot_list = [\n ('max_iter', 'Train score'),\n ('max_iter', 'Test score'),\n ('max_iter', 'n_iter_'),\n ('max_iter', 'Fit time (sec)'),\n]\n\nnrows = 2\nncols = int(np.ceil(len(plot_list) / 2.))\nfig, axes = plt.subplots(nrows=nrows, ncols=ncols, figsize=(6 * ncols,\n 4 * nrows))\naxes[0, 0].get_shared_y_axes().join(axes[0, 0], axes[0, 1])\n\nfor ax, (x_axis, y_axis) in zip(axes.ravel(), plot_list):\n for criterion, group_df in results_df.groupby(lines):\n group_df.plot(x=x_axis, y=y_axis, label=criterion, ax=ax)\n ax.set_title(y_axis)\n ax.legend(title=lines)\n\nfig.tight_layout()\nplt.show()"

"# Authors: Tom Dupre la Tour\n#\n# License: BSD 3 clause\nimport time\nimport sys\n\nimport pandas as pd\nimport numpy as np\nimport matplotlib.pyplot as plt\n\nfrom sklearn import linear_model\nfrom sklearn.datasets import fetch_openml\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.utils.testing import ignore_warnings\nfrom sklearn.exceptions import ConvergenceWarning\nfrom sklearn.utils import shuffle\n\nprint(__doc__)\n\n\ndef load_mnist(n_samples=None, class_0='0', class_1='8'):\n \"\"\"Load MNIST, select two classes, shuffle and return only n_samples.\"\"\"\n # Load data from http://openml.org/d/554\n mnist = fetch_openml('mnist_784', version=1)\n\n # take only two classes for binary classification\n mask = np.logical_or(mnist.target == class_0, mnist.target == class_1)\n\n X, y = shuffle(mnist.data[mask], mnist.target[mask], random_state=42)\n if n_samples is not None:\n X, y = X[:n_samples], y[:n_samples]\n return X, y\n\n\n@ignore_warnings(category=ConvergenceWarning)\ndef fit_and_score(estimator, max_iter, X_train, X_test, y_train, y_test):\n \"\"\"Fit the estimator on the train set and score it on both sets\"\"\"\n estimator.set_params(max_iter=max_iter)\n estimator.set_params(random_state=0)\n\n start = time.time()\n estimator.fit(X_train, y_train)\n\n fit_time = time.time() - start\n n_iter = estimator.n_iter_\n train_score = estimator.score(X_train, y_train)\n test_score = estimator.score(X_test, y_test)\n\n return fit_time, n_iter, train_score, test_score\n\n\n# Define the estimators to compare\nestimator_dict = {\n 'No stopping criterion':\n linear_model.SGDClassifier(n_iter_no_change=3),\n 'Training loss':\n linear_model.SGDClassifier(early_stopping=False, n_iter_no_change=3,\n tol=0.1),\n 'Validation score':\n linear_model.SGDClassifier(early_stopping=True, n_iter_no_change=3,\n tol=0.0001, validation_fraction=0.2)\n}\n\n# Load the dataset\nX, y = load_mnist(n_samples=10000)\nX_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.5,\n random_state=0)\n\nresults = []\nfor estimator_name, estimator in estimator_dict.items():\n print(estimator_name + ': ', end='')\n for max_iter in range(1, 50):\n print('.', end='')\n sys.stdout.flush()\n\n fit_time, n_iter, train_score, test_score = fit_and_score(\n estimator, max_iter, X_train, X_test, y_train, y_test)\n\n results.append((estimator_name, max_iter, fit_time, n_iter,\n train_score, test_score))\n print('')\n\n# Transform the results in a pandas dataframe for easy plotting\ncolumns = [\n 'Stopping criterion', 'max_iter', 'Fit time (sec)', 'n_iter_',\n 'Train score', 'Test score'\n]\nresults_df = pd.DataFrame(results, columns=columns)\n\n# Define what to plot (x_axis, y_axis)\nlines = 'Stopping criterion'\nplot_list = [\n ('max_iter', 'Train score'),\n ('max_iter', 'Test score'),\n ('max_iter', 'n_iter_'),\n ('max_iter', 'Fit time (sec)'),\n]\n\nnrows = 2\nncols = int(np.ceil(len(plot_list) / 2.))\nfig, axes = plt.subplots(nrows=nrows, ncols=ncols, figsize=(6 * ncols,\n 4 * nrows))\naxes[0, 0].get_shared_y_axes().join(axes[0, 0], axes[0, 1])\n\nfor ax, (x_axis, y_axis) in zip(axes.ravel(), plot_list):\n for criterion, group_df in results_df.groupby(lines):\n group_df.plot(x=x_axis, y=y_axis, label=criterion, ax=ax)\n ax.set_title(y_axis)\n ax.legend(title=lines)\n\nfig.tight_layout()\nplt.show()"

]

}

],