{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "# Exploring Label Relations\n", "\n", "Multi-label classification tends to have problems with overfitting and underfitting classifiers when the label space is large, especially in problem transformation approaches. A well known approach to remedy this is to split the problem into subproblems with smaller label subsets to improve the generalization quality. \n", "\n", "Scikit-multilearn library is the first Python library to provide this functionality, this will guide your through using different libraries for label space division. Let's start with loading up the well-cited ``emotions`` dataset, that use throughout the User Guide:" ] }, { "cell_type": "code", "execution_count": 1, "metadata": { "scrolled": true }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "emotions:train - exists, not redownloading\n", "emotions:test - exists, not redownloading\n" ] } ], "source": [ "from skmultilearn.dataset import load_dataset\n", "X_train, y_train, feature_names, label_names = load_dataset('emotions', 'train')\n", "X_test, y_test, _, _ = load_dataset('emotions', 'test')" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Label relationships can be exploited in a handful of ways:\n", "\n", "1. inferring the label space division from the label assignment matrix in the training set:\n", " - through building a label graph and [inferring community structure of this graph](http://www.mdpi.com/1099-4300/18/8/282/htm), this can be facilitated with three network libraries in scikit-multilearn: NetworkX (BSD), igraph (GPL) and graphtool (GPL)\n", " - through using a traditional clustering approach from scikit-learn to cluster label assignment vectors, ex. using k-means, this usually required parameter estimation\n", "2. employing expert knowledge to divide the label space\n", "3. random label space partitioning with methods like [random k-label sets](https://ieeexplore.ieee.org/document/5567103/)\n", "\n", "\n", "In most cases these approaches are used with a Label Powerset problem transformation classifier and a base multi-class classifier, for the examples in this chapter we will use sklearn's Gaussian Naive Bayes classifier, but you can use whatever classifiers you in your ensembles.\n", "\n", "Let's go through the approaches:" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Detecting communities in Label Relations Graph \n", "\n", "Exploring label relations using the current methods of Network Science is a new approach to improve classification results. This area is still under research, both in terms of methods used for label space division and in terms of what qualities should be represented in the Label Relations Graph. \n", "\n", "In scikit-multilearn classifying with label space division based on label graphs requires three elements:\n", "\n", "- selecting a graph builder, a class that constructs a graph based on the label assignment matrix `y`, at the moment scikit-multilearn provides one such graph builder, based on the notion of label co-occurrence\n", "\n", "- selecting a Label Graph clusterer which employs community detection methods from different sources to provide a label space clustering\n", "\n", "- selecting a classification approach, i.e. how to train and merge results of classifiers, scikit-multilearn provides two approaches: \n", "\n", " - a partitioning classifier which trains a classifier per label cluster, assuming they are disjoint, and merges the results of each subclassifier's prediction\n", " - a majority voting classifier that trains a classifier per label clusters, but if they overlap, it follows the decision of the majority of subclassifiers concerning assigning the label or not\n", " \n", "Let's start with looking at the Label Graph builder.\n", "\n", "\n", "\n", "### Building a Label Graph" ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [], "source": [ "from skmultilearn.cluster import LabelCooccurrenceGraphBuilder" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "This graph builder constructs a Label Graph based on the output matrix where two label nodes are connected when at least one sample is labeled with both of them. If the graph is weighted, the weight of an edge between two label nodes is the number of samples labeled with these two labels. Self-edge weights contain the number of samples with a given label." ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [], "source": [ "graph_builder = LabelCooccurrenceGraphBuilder(weighted=True, include_self_edges=False)" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "6 labels, 14 edges\n", "{(1, 2): 58.0, (0, 1): 33.0, (1, 3): 6.0, (4, 5): 9.0, (1, 4): 1.0, (0, 2): 9.0, (1, 5): 6.0, (0, 5): 61.0, (0, 4): 4.0, (2, 3): 66.0, (2, 5): 5.0, (3, 4): 56.0, (2, 4): 60.0, (3, 5): 2.0}\n" ] } ], "source": [ "edge_map = graph_builder.transform(y_train)\n", "print(\"{} labels, {} edges\".format(len(label_names), len(edge_map)))\n", "print(edge_map)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The dictionary ``edge_map`` contains the adjacency matrix in dictionary-of-keys format, each key is a label number tuple, weight is the number of samples with the two labels assigned. Its values will be used by all of the supported Label Graph Clusterers below:\n", "\n", "- NetworkX\n", "- igraph\n", "- graph-tool\n", "\n", "All these clusterers take their names from the respected Python graph/network libraries which they are using to infer community structure and provide the label space clustering." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### NetworkX" ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [], "source": [ "from skmultilearn.cluster import NetworkXLabelGraphClusterer\n", "\n", "# we define a helper function for visualization purposes\n", "def to_membership_vector(partition):\n", " return { \n", " member : partition_id\n", " for partition_id, members in enumerate(partition) \n", " for member in members\n", " }\n" ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [], "source": [ "clusterer = NetworkXLabelGraphClusterer(graph_builder, method='louvain')" ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[0, 1, 5],\n", " [2, 3, 4]])" ] }, "execution_count": 7, "metadata": {}, "output_type": "execute_result" } ], "source": [ "partition = clusterer.fit_predict(X_train,y_train)\n", "partition" ] }, { "cell_type": "code", "execution_count": 8, "metadata": {}, "outputs": [], "source": [ "membership_vector = to_membership_vector(partition)" ] }, { "cell_type": "code", "execution_count": 9, "metadata": {}, "outputs": [], "source": [ "import networkx as nx\n", "names_dict = dict(enumerate(x[0].replace('-','-\\n') for x in label_names))" ] }, { "cell_type": "code", "execution_count": 10, "metadata": {}, "outputs": [], "source": [ "import matplotlib.pyplot as plt\n", "%matplotlib inline" ] }, { "cell_type": "code", "execution_count": 11, "metadata": {}, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "nx.draw(\n", " clusterer.graph_,\n", " pos=nx.circular_layout(clusterer.graph_),\n", " labels=names_dict, \n", " with_labels = True,\n", " width = [10*x/y_train.shape[0] for x in clusterer.weights_['weight']],\n", " node_color = [membership_vector[i] for i in range(y_train.shape[1])],\n", " cmap=plt.cm.Spectral,\n", " node_size=100,\n", " font_size=14\n", ")" ] }, { "cell_type": "code", "execution_count": 12, "metadata": {}, "outputs": [], "source": [ "from skmultilearn.ensemble import LabelSpacePartitioningClassifier\n", "from skmultilearn.problem_transform import LabelPowerset\n", "from sklearn.naive_bayes import GaussianNB\n", "from sklearn.metrics import accuracy_score" ] }, { "cell_type": "code", "execution_count": 13, "metadata": {}, "outputs": [], "source": [ "classifier = LabelSpacePartitioningClassifier(\n", " classifier = LabelPowerset(classifier=GaussianNB()),\n", " clusterer = clusterer\n", ")" ] }, { "cell_type": "code", "execution_count": 14, "metadata": {}, "outputs": [], "source": [ "classifier.fit(X_train, y_train)\n", "prediction = classifier.predict(X_test)" ] }, { "cell_type": "code", "execution_count": 15, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.17821782178217821" ] }, "execution_count": 15, "metadata": {}, "output_type": "execute_result" } ], "source": [ "accuracy_score(y_test, prediction)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Using iGraph" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "\n", "\n", "To use igraph with scikit-multilearn you need to install the igraph python package:\n", "```bash\n", "$ pip install python-igraph\n", "```\n", "\n", "Do not install the ``igraph`` package which is not the correct python-igraph library. Information about build requirements of ``python-igraph`` can be found in the [library documentation](http://igraph.org/python/#pyinstall). \n", "\n", "Let's load the python igraph library and scikit-multilearn's igraph-based clusterer." ] }, { "cell_type": "code", "execution_count": 16, "metadata": {}, "outputs": [], "source": [ "from skmultilearn.cluster import IGraphLabelGraphClusterer\n", "import igraph as ig" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Igraph provides a set of community detection methods, out of which the following are supported:\n", "\n", "\n", "| Method name string | Description |\n", "|--------------------|-------------|\n", "| ``fastgreedy`` | Detecting communities with largest modularity using incremental greedy search |\n", "| ``infomap`` | Detecting communities through information flow compressing simulated via random walks |\n", "| ``label_propagation`` | Detecting communities from colorings via multiple label propagation on the graph |\n", "| ``leading_eigenvector`` | Detecting communities with largest modularity through adjacency matrix eigenvectors |\n", "| ``multilevel`` | Recursive communitiy detection with largest modularity step by step maximization |\n", "| ``walktrap`` | Finding communities by trapping many random walks |\n" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Each of them denotes a ``community_*`` method of the Graph object, you can read more about the methods in [igraph documentation](http://igraph.org/python/doc/igraph.Graph-class.html#community_fastgreedy) and in comparison of their performance in [multi-label classification](http://www.mdpi.com/1099-4300/18/8/282/htm).\n", "\n", "Let's start with detecting a community structure in the label co-occurrence graph and visualizing it with igraph." ] }, { "cell_type": "code", "execution_count": 17, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[0, 5], [1, 2, 3, 4]], dtype=object)" ] }, "execution_count": 17, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clusterer_igraph = IGraphLabelGraphClusterer(graph_builder=graph_builder, method='walktrap')\n", "partition = clusterer_igraph.fit_predict(X_train, y_train)\n", "partition" ] }, { "cell_type": "code", "execution_count": 18, "metadata": {}, "outputs": [ { "data": { "image/svg+xml": [ "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "\n", "\n", "\n" ], "text/plain": [ "" ] }, "execution_count": 18, "metadata": {}, "output_type": "execute_result" } ], "source": [ "\n", "colors = ['red', 'white', 'blue']\n", "membership_vector = to_membership_vector(partition)\n", "visual_style = {\n", " \"vertex_size\" : 20,\n", " \"vertex_label\": [x[0] for x in label_names],\n", " \"edge_width\" : [10*x/y_train.shape[0] for x in clusterer_igraph.graph_.es['weight']],\n", " \"vertex_color\": [colors[membership_vector[i]] for i in range(y_train.shape[1])],\n", " \"bbox\": (400,400),\n", " \"margin\": 80,\n", " \"layout\": clusterer_igraph.graph_.layout_circle()\n", " \n", "}\n", "\n", "ig.plot(clusterer_igraph.graph_, **visual_style)" ] }, { "cell_type": "code", "execution_count": 19, "metadata": {}, "outputs": [], "source": [ "classifier = LabelSpacePartitioningClassifier(\n", " classifier = LabelPowerset(classifier=GaussianNB()),\n", " clusterer = clusterer_igraph\n", ")\n", "classifier.fit(X_train, y_train)\n", "prediction = classifier.predict(X_test)" ] }, { "cell_type": "code", "execution_count": 20, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.19306930693069307" ] }, "execution_count": 20, "metadata": {}, "output_type": "execute_result" } ], "source": [ "accuracy_score(y_test, prediction)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Stochastic Blockmodel from graph-tool" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Another approach to label space division is to fit a [Stochastic Block Model](https://en.wikipedia.org/wiki/Stochastic_block_model) to the label graph. An efficient implementation of the Stochastic Block Model in Python is provided by [graphtool](https://graph-tool.skewed.de). Note that using graphtool incurs GPL requirements on your code." ] }, { "cell_type": "code", "execution_count": 21, "metadata": {}, "outputs": [], "source": [ "from skmultilearn.cluster.graphtool import GraphToolLabelGraphClusterer, StochasticBlockModel" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The `StochasticBlockModel` class fits the model and specifies the variant of SBM to be used, it can include:\n", "\n", "- whether to use a nested blockmodel or not\n", "- whether to take degree correlation into account\n", "- whether to allow overlapping communities\n", "- how to model weights of label relationships\n", "\n", "Selecting these parameters efficiently for multi-label purposes is still researched, but reading the [inference documentation](https://graph-tool.skewed.de/static/doc/inference.html) in graphtool will give you an intuition what to choose.\n", "\n", "As the emotions data set is small there is no reason to use the nested model, we select the real-normal weight model as it is reasonable to believe that label assignments come from an i.i.d source and should follow some limit theorem." ] }, { "cell_type": "code", "execution_count": 22, "metadata": {}, "outputs": [], "source": [ "model = StochasticBlockModel(nested=False, use_degree_correlation=True, allow_overlap=False, weight_model='real-normal')" ] }, { "cell_type": "code", "execution_count": 23, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[0, 1, 5],\n", " [2, 3, 4]])" ] }, "execution_count": 23, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clusterer_graphtool = GraphToolLabelGraphClusterer(graph_builder=graph_builder, model=model)\n", "clusterer_graphtool.fit_predict(None, y_train)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The above partition was generated by the model, let's visualize it." ] }, { "cell_type": "code", "execution_count": 28, "metadata": {}, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/plain": [ "', for Graph 0x7fd284397b90, at 0x7fd283cf1450>" ] }, "execution_count": 28, "metadata": {}, "output_type": "execute_result" } ], "source": [ "node_label = clusterer_graphtool.graph_.new_vertex_property(\"string\")\n", "\n", "for i, v in enumerate(clusterer_graphtool.graph_.vertices()):\n", " node_label[v] = label_names[i][0]\n", " \n", "clusterer_graphtool.model.model_.draw(vertex_text=node_label)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We can use this clusterer as an argument for the label space partitioning classifier, as we did not enable overlapping communities:" ] }, { "cell_type": "code", "execution_count": 29, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.19306930693069307" ] }, "execution_count": 29, "metadata": {}, "output_type": "execute_result" } ], "source": [ "classifier = LabelSpacePartitioningClassifier(\n", " classifier = LabelPowerset(classifier=GaussianNB()),\n", " clusterer = clusterer_graphtool\n", ")\n", "classifier.fit(X_train, y_train)\n", "prediction = classifier.predict(X_test)\n", "accuracy_score(y_test, prediction)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Now let's try to go with the same variant of the model, but now we allow overlapping communities:" ] }, { "cell_type": "code", "execution_count": 30, "metadata": {}, "outputs": [], "source": [ "model = StochasticBlockModel(nested=False, use_degree_correlation=True, allow_overlap=True, weight_model='real-normal')" ] }, { "cell_type": "code", "execution_count": 55, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[0, 1, 2, 5],\n", " [2, 3, 4, 5]])" ] }, "execution_count": 55, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clusterer_graphtool = GraphToolLabelGraphClusterer(graph_builder=graph_builder, model=model)\n", "clusterer_graphtool.fit_predict(None, y_train)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We have a division, note that we train the same number of classifiers as in the partitioning case. Let's visualize label membership likelihoods alongside the division:" ] }, { "cell_type": "code", "execution_count": 56, "metadata": {}, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/plain": [ "', for Graph 0x7fd283d14bd0, at 0x7fd259aa2a90>" ] }, "execution_count": 56, "metadata": {}, "output_type": "execute_result" } ], "source": [ "node_label = clusterer_graphtool.graph_.new_vertex_property(\"string\")\n", "\n", "for i, v in enumerate(clusterer_graphtool.graph_.vertices()):\n", " node_label[v] = label_names[i][0]\n", "\n", "clusterer_graphtool.model.model_.draw(vertex_text=node_label, vertex_text_color='black')" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We can now perform classification, but for it to work we now need to use a classifier that can decide whether to assign a label if more than one subclassifiers were making a decision about the label. We will use the ``MajorityVotingClassifier`` which makes a decision if the majority of classifiers decide to assign the label." ] }, { "cell_type": "code", "execution_count": 57, "metadata": {}, "outputs": [], "source": [ "from skmultilearn.ensemble.voting import MajorityVotingClassifier" ] }, { "cell_type": "code", "execution_count": 58, "metadata": {}, "outputs": [], "source": [ "classifier = MajorityVotingClassifier(\n", " classifier=LabelPowerset(classifier=GaussianNB()),\n", " clusterer=clusterer_graphtool\n", ")\n", "classifier.fit(X_train, y_train)\n", "prediction = classifier.predict(X_test)" ] }, { "cell_type": "code", "execution_count": 59, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.25742574257425743" ] }, "execution_count": 59, "metadata": {}, "output_type": "execute_result" } ], "source": [ "accuracy_score(y_test, prediction)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Using scikit-learn clusterers " ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Scikit-learn offers a variety of [clustering](http://scikit-learn.org/stable/modules/clustering.html) methods, some of which have been applied to dividing the label space into subspaces in multi-label classification. The main problem which often concerns these approaches is the need to empirically fit the parameter of the number of clusters to select. \n", "\n", "scikit-multilearn provides a clusterer which does not build a graph, instead it employs the scikit-multilearn clusterer on transposed label assignment vectors, i.e. a vector for a given label is a vector of all samples' assignment values. To use this approach, just import a scikit-learn cluster, and pass its instance as a parameter." ] }, { "cell_type": "code", "execution_count": 36, "metadata": {}, "outputs": [], "source": [ "from skmultilearn.cluster import MatrixLabelSpaceClusterer\n", "from sklearn.cluster import KMeans" ] }, { "cell_type": "code", "execution_count": 37, "metadata": {}, "outputs": [], "source": [ "matrix_clusterer = MatrixLabelSpaceClusterer(clusterer=KMeans(n_clusters=2))" ] }, { "cell_type": "code", "execution_count": 38, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[2, 3, 4],\n", " [0, 1, 5]])" ] }, "execution_count": 38, "metadata": {}, "output_type": "execute_result" } ], "source": [ "matrix_clusterer.fit_predict(X_train, y_train)" ] }, { "cell_type": "code", "execution_count": 39, "metadata": {}, "outputs": [], "source": [ "classifier = LabelSpacePartitioningClassifier(\n", " classifier = LabelPowerset(classifier=GaussianNB()),\n", " clusterer = matrix_clusterer\n", ")" ] }, { "cell_type": "code", "execution_count": 40, "metadata": {}, "outputs": [], "source": [ "classifier.fit(X_train, y_train)\n", "prediction = classifier.predict(X_test)" ] }, { "cell_type": "code", "execution_count": 41, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.17821782178217821" ] }, "execution_count": 41, "metadata": {}, "output_type": "execute_result" } ], "source": [ "accuracy_score(y_test, prediction)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Fixed partition based on expert knowledge" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "There may be cases where we know something about the label relationships based on expert or intuitive knowledge, or perhaps our knowledge comes from a different machine learning model, or it is crowdsourced, in all of these cases, scikit-multilearn let's you use this knowledge to your advantage. Let's see this on our exampel data set. It has six labels that denote emotions:" ] }, { "cell_type": "code", "execution_count": 42, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "[(u'amazed-suprised', [u'0', u'1']),\n", " (u'happy-pleased', [u'0', u'1']),\n", " (u'relaxing-calm', [u'0', u'1']),\n", " (u'quiet-still', [u'0', u'1']),\n", " (u'sad-lonely', [u'0', u'1']),\n", " (u'angry-aggresive', [u'0', u'1'])]" ] }, "execution_count": 42, "metadata": {}, "output_type": "execute_result" } ], "source": [ "label_names" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Looking at label names we might see, that labels `quiet-still` and `angry-agressive` are contradictory, but one can be `amazed` both in the `happy/relaxing` context, in the `sad/agresive` context. Also one can be easily `pleased/relaxed` and/or `calm` but not actually amazed. We thus come up with a new intuitive label space division:" ] }, { "cell_type": "code", "execution_count": 43, "metadata": {}, "outputs": [], "source": [ "from skmultilearn.ensemble import MajorityVotingClassifier\n", "from skmultilearn.cluster import FixedLabelSpaceClusterer\n", "from skmultilearn.problem_transform import LabelPowerset\n", "from sklearn.ensemble import RandomForestClassifier\n", "\n", "classifier = MajorityVotingClassifier(\n", " classifier = LabelPowerset(\n", " classifier=RandomForestClassifier(n_estimators=100),\n", " require_dense = [False, True]\n", " ),\n", " require_dense = [True, True],\n", " clusterer = FixedLabelSpaceClusterer(clusters=[[0,1, 2], [2, 3 ,4], [0, 4, 5]])\n", ")\n", "\n", "# train\n", "classifier.fit(X_train, y_train)\n", "\n", "# predict\n", "predictions = classifier.predict(X_test)" ] }, { "cell_type": "code", "execution_count": 44, "metadata": { "scrolled": false }, "outputs": [ { "data": { "text/plain": [ "0.29702970297029702" ] }, "execution_count": 44, "metadata": {}, "output_type": "execute_result" } ], "source": [ "accuracy_score(y_test, predictions)" ] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 2 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython2", "version": "2.7.14" } }, "nbformat": 4, "nbformat_minor": 2 }