Skip to content

Instantly share code, notes, and snippets.

Show Gist options
  • Save luanjunyi/6632d4c0f92bc30750f4 to your computer and use it in GitHub Desktop.
Save luanjunyi/6632d4c0f92bc30750f4 to your computer and use it in GitHub Desktop.
Kaggle sklearn competition solution(ipython notebook)
Display the source blob
Display the rendered blob
Raw
{
"metadata": {
"name": ""
},
"nbformat": 3,
"nbformat_minor": 0,
"worksheets": [
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"** Spoiler: ** This solution can get 99% in the public leaderboard, so continue reading will deprive you of pretty much all the fun of solving the puzzle yourselves.\n",
"\n",
"## General idea\n",
"\n",
"This is not hard data. Using default parametered RBF kernel SVM will get 92%. The discussions in the forum are mainly about how to use feature engineering to enhance SVM. Using well tuned PCA will get 94%. One tutorial used Sparse Filtering and got 96%. I'm never a big fan of neural networks or deep learning and found out that modeling the data is the key. By finding out the parameters of the distribution, we are able to transform each record into the latent variable in the distribution. Using the latent variables alone as features significantly outperforms the original data or PCA components.\n",
"\n",
"The treatment consists of three steps. Firstly, I used run whitened PCA on the orignial data. From there I plotted KDE(Kernal Density Estimation) figure for each of the PCA components. By observing the patterns in the plots, GMM(Gaussian Mixture Model) is one of the plausible distribution. The second step is therefore fitting GMM on the data. With the trained GMM, we can get for each record the belongingness to each GMM components. Lastly, we use the belongingness as features and feed them to SVM.\n",
"\n",
"In this case, hyper parameter tunning makes a huge difference. The key parameters are PCA's components number, PCA's whittening or not, GMM's components number and GMM's covariance type.\n",
"\n",
"Another tip is to include the testing data in the unsupervised learning phases(PCA and GMM). Since there are 9000 testing data and only 1000 training data, it makes sense to take advantage of the testing set. This tip improved my score from 98% to 99%.\n",
"\n",
"## First thing first: loading data"
]
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"import numpy as np\n",
"import pandas as pd\n",
"from sklearn.decomposition.pca import PCA\n",
"from sklearn.grid_search import GridSearchCV\n",
"from sklearn.svm import SVC\n",
"from sklearn.mixture import GMM\n",
"from sklearn.base import BaseEstimator\n",
"import matplotlib.pyplot as plt\n",
"\n",
"X_test = pd.read_csv('test.csv', header=None).as_matrix()\n",
"y = pd.read_csv('trainLabels.csv', header=None)[0].as_matrix()\n",
"X = pd.read_csv('train.csv', header=None).as_matrix()"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 6
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The training data is 1000 by 40 and the testing data is 9000 by 40. Let's reduce the demenstion to 2 so that the training data can be plotted. As we can see from the plot below, the 2D graph is not much separable."
]
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"pca2 = PCA(n_components=2, whiten=True)\n",
"pca2.fit(np.r_[X, X_test])\n",
"X_pca = pca2.transform(X)\n",
"i0 = np.argwhere(y == 0)[:, 0]\n",
"i1 = np.argwhere(y == 1)[:, 0]\n",
"X0 = X_pca[i0, :]\n",
"X1 = X_pca[i1, :]\n",
"plt.plot(X0[:, 0], X0[:, 1], 'ro')\n",
"plt.plot(X1[:, 0], X1[:, 1], 'b*')"
],
"language": "python",
"metadata": {},
"outputs": [
{
"output_type": "stream",
"stream": "stdout",
"text": [
"(490, 2)\n"
]
},
{
"metadata": {},
"output_type": "pyout",
"prompt_number": 11,
"text": [
"[<matplotlib.lines.Line2D at 0x111fda550>]"
]
},
{
"metadata": {},
"output_type": "display_data",
"png": "iVBORw0KGgoAAAANSUhEUgAAAXAAAAD9CAYAAAClQCyNAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nUzTACSFEp2iAE0KITtEAJ4QQnfoCyTV0mgHSTQoAAAAASUVORK5CYII=\n",
"text": [
"<matplotlib.figure.Figure at 0x111f9cfd0>"
]
}
],
"prompt_number": 11
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"##PCA\n",
"Since we have no idea about the meaning of any features, we loss no information running PCA on the data. Many have found out that using the first 12 PCA components as feature will improve SVM accuracy from 92% to 94%. It seem that the last three components are nosie and should be discarded. "
]
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"pca = PCA(whiten=True)\n",
"X_all = pca.fit_transform(np.r_[X, X_test])\n",
"print pca.explained_variance_ratio_"
],
"language": "python",
"metadata": {},
"outputs": [
{
"output_type": "stream",
"stream": "stdout",
"text": [
"[ 2.61338893e-01 2.03953972e-01 7.91104009e-02 4.81480076e-02\n",
" 4.54167442e-02 4.42563499e-02 4.04285747e-02 3.03524876e-02\n",
" 2.35202947e-02 1.92164610e-02 1.61483800e-02 1.25565492e-02\n",
" 7.65669101e-03 7.59267345e-03 7.57067587e-03 7.45196972e-03\n",
" 7.39437803e-03 7.33602203e-03 7.32154020e-03 7.26446366e-03\n",
" 7.24687802e-03 7.13785979e-03 7.08554112e-03 7.05759361e-03\n",
" 7.01144882e-03 6.98357997e-03 6.90865448e-03 6.88494883e-03\n",
" 6.84557325e-03 6.83367011e-03 6.74339150e-03 6.70112785e-03\n",
" 6.63533428e-03 6.57374179e-03 6.49188079e-03 6.46379503e-03\n",
" 6.35945185e-03 2.13086486e-31 3.58428641e-32 1.30173625e-32]\n"
]
}
],
"prompt_number": 13
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"##KDE plots and QQ plots\n",
"\n",
"The next sensible thing to do is the see what the distributions of each PCA components look like. We can plot histograms, but KDE with Gaussian kernal works better here. To save you some time, I only draw 4 of them. In fact, components 1-37 are all similar and the last three noise components are different."
]
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"def kde_plot(x):\n",
" from scipy.stats.kde import gaussian_kde\n",
" kde = gaussian_kde(x)\n",
" positions = np.linspace(x.min(), x.max())\n",
" smoothed = kde(positions)\n",
" plt.plot(positions, smoothed)\n",
"\n",
"def qq_plot(x):\n",
" from scipy.stats import probplot\n",
" probplot(x, dist='norm', plot=plt)\n",
" "
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 20
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"kde_plot(X_all[:, 0])"
],
"language": "python",
"metadata": {},
"outputs": [
{
"metadata": {},
"output_type": "display_data",
"png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAAD9CAYAAAC2l2x5AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRK5CYII=\n",
"text": [
"<matplotlib.figure.Figure at 0x111f950d0>"
]
}
],
"prompt_number": 15
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"kde_plot(X_all[:, 2])\n"
],
"language": "python",
"metadata": {},
"outputs": [
{
"metadata": {},
"output_type": "display_data",
"png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAAD9CAYAAAC2l2x5AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nrkJggg==\n",
"text": [
"<matplotlib.figure.Figure at 0x11203d950>"
]
}
],
"prompt_number": 17
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"kde_plot(X_all[:, 30])\n"
],
"language": "python",
"metadata": {},
"outputs": [
{
"metadata": {},
"output_type": "display_data",
"png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAAD9CAYAAAC2l2x5AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIp36/z2zk/yzW7ntAAAAAElFTkSuQmCC\n",
"text": [
"<matplotlib.figure.Figure at 0x111d152d0>"
]
}
],
"prompt_number": 18
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"kde_plot(X_all[:, 38])\n"
],
"language": "python",
"metadata": {},
"outputs": [
{
"metadata": {},
"output_type": "display_data",
"png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAD9CAYAAABDaefJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngAJCqANAAfk/VHPNZfaasL4AAAAASUVORK5CYII=\n",
"text": [
"<matplotlib.figure.Figure at 0x112025c90>"
]
}
],
"prompt_number": 19
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"So are they Gaussian? I tried to model the data using multi-variant Gaussian but didn't get better result than simple SVM. Let's see the QQ plot assuming the data is from Gaussian distribution."
]
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"qq_plot(X[:,0])"
],
"language": "python",
"metadata": {},
"outputs": [
{
"metadata": {},
"output_type": "display_data",
"png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAEWCAYAAACKSkfIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
"text": [
"<matplotlib.figure.Figure at 0x111f90a10>"
]
}
],
"prompt_number": 22
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"qq_plot(X[:,2])"
],
"language": "python",
"metadata": {},
"outputs": [
{
"metadata": {},
"output_type": "display_data",
"png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAEWCAYAAACKSkfIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n0+dju8zaAAAAAElFTkSuQmCC\n",
"text": [
"<matplotlib.figure.Figure at 0x111fd7290>"
]
}
],
"prompt_number": 23
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"qq_plot(X[:,38])"
],
"language": "python",
"metadata": {},
"outputs": [
{
"metadata": {},
"output_type": "display_data",
"png": "iVBORw0KGgoAAAANSUhEUgAAAYQAAAEWCAYAAABmE+CbAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/H8AffXRAHG+cgAAAABJRU5ErkJggg==\n",
"text": [
"<matplotlib.figure.Figure at 0x111fbd350>"
]
}
],
"prompt_number": 24
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"To save you the time, all components including the noises, have one common pattern in the QQ plots: the middle range fits well with Gaussian but the extreme quantiles drift away. One plausible distribution that fit both the KDE plots and QQ plots is GMM. Unlike multi-variant Gaussian, data from GMM have one sample specific parameter that dipict the belongingness of that sample to each of the GMM components. This latent vector of belongingness is more of merit than the PCA components since it determines the value of the components under the GMM assumption.\n",
"\n",
"The idea is to use the belongingness alone as features to feed to SVM. There will be lots of grid search to get optimal numboer of PCA components, GMM components and GMM covarance type. The idea is implemented in below code. You need to inherite from BaseEstimator in order to work with other sklearn facility like GridSearchCV."
]
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"class PcaGmm(BaseEstimator):\n",
" def __init__(self, X_all,\n",
" pca_components = 12, gmm_components = 4,\n",
" covariance_type = \"full\", min_covar = 0.1,\n",
" gamma = 0, C = 1.0):\n",
"\n",
" self.pca_components = pca_components\n",
" self.gmm_components = gmm_components\n",
" self.covariance_type = covariance_type\n",
" self.min_covar = min_covar\n",
" self.gamma = gamma\n",
" self.C = C\n",
" self.X_all = X_all\n",
" X_all = X_all[:, :pca_components]\n",
" self.gmm = GMM(n_components = gmm_components,\n",
" covariance_type = covariance_type,\n",
" min_covar = min_covar)\n",
" self.gmm.fit(X_all)\n",
"\n",
" def fit(self, X, y):\n",
" X = X[:, :self.pca_components]\n",
" X = self.gmm.predict_proba(X)\n",
" self.svm = SVC(C = self.C, gamma = self.gamma)\n",
" self.svm.fit(X, y)\n",
"\n",
" def predict(self, X):\n",
" X = X[:, :self.pca_components]\n",
" return self.svm.predict(self.gmm.predict_proba(X))\n",
"\n",
" def score(self, X, y):\n",
" y_pred = self.predict(X)\n",
" return accuracy_score(y, y_pred)\n",
"\n",
" def transform(self, X, y = None):\n",
" X = X[:, :self.pca_components]\n",
" return self.gmm.predict_proba(X)\n",
"\n",
" def __str__(self):\n",
" return \"PCA(%d)-GMM(%d, %s, %f)-SVM(C=%f, gamma=%f)\" % (self.pca_components, self.gmm_components,\n",
" self.covariance_type, self.min_covar,\n",
" self.C, self.gamma)\n"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 28
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Careful grid search will find the key hyper parameters as below:\n",
"\n",
"* PCA components no. is 12\n",
"* GMM components no. is 4\n",
"* GMM covarance type is 'full'\n",
"\n",
"My model got 0.99143 in the public leaderboard. Let's see the PCA and GMM generated feature compare to the original one."
]
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"clf = PcaGmm(X_all,\n",
" 12, 4, 'full', 0, gamma = .6, C = 0.3)\n",
"X_t = clf.transform(pca.transform(X))\n",
"X0 = X_t[i0, :]\n",
"X1 = X_t[i1, :]\n",
"pca2 = PCA(n_components=2)\n",
"pca2.fit(np.r_[X0, X1])\n",
"X0 = pca2.transform(X0)\n",
"X1 = pca2.transform(X1)\n",
"plt.plot(X0[:, 0], X0[:, 1], 'ro')\n",
"plt.plot(X1[:, 0], X1[:, 1], 'b*')"
],
"language": "python",
"metadata": {},
"outputs": [
{
"metadata": {},
"output_type": "pyout",
"prompt_number": 31,
"text": [
"[<matplotlib.lines.Line2D at 0x11202d9d0>]"
]
},
{
"metadata": {},
"output_type": "display_data",
"png": "iVBORw0KGgoAAAANSUhEUgAAAX8AAAD9CAYAAABUS3cAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYkEKfxERC1L4i4hYkMJfRMSC/j9m6yDpaOAjywAAAABJRU5ErkJggg==\n",
"text": [
"<matplotlib.figure.Figure at 0x111fda4d0>"
]
}
],
"prompt_number": 31
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Looks much better, huh?"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Acknowlegement\n",
"\n",
"I can't figure it out without help from [Peter Williams](http://www.linkedin.com/profile/view?id=57878071&authType=OUT_OF_NETWORK&authToken=-mnJ&locale=en_US&trk=tyah2&trkInfo=tarId%3A1402026200787%2Ctas%3Apeter%20wi%2Cidx%3A1-1-1). Peter was on top of the leaderboard when I got there. He patiently answered lots of my silly questions via Email. We have a [thread](http://www.kaggle.com/c/data-science-london-scikit-learn/forums/t/8104/anyone-in-the-99-league-care-to-share-the-solution) at Kaggle forum, thank all for participating the discussion and bearing with my unfamiliar with statistical modeling. "
]
},
{
"cell_type": "code",
"collapsed": false,
"input": [],
"language": "python",
"metadata": {},
"outputs": []
}
],
"metadata": {}
}
]
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment