Skip to content

Instantly share code, notes, and snippets.

@jelena-markovic

jelena-markovic/HIV3TC_example.ipynb

Created August 11, 2017 01:29
Show Gist options
  • Save jelena-markovic/0be464412f734ecac97514663893a8a4 to your computer and use it in GitHub Desktop.
Save jelena-markovic/0be464412f734ecac97514663893a8a4 to your computer and use it in GitHub Desktop.
Download ZIP
St.Louis
Display the source blob
Display the rendered blob
Raw
HIV3TC_example.ipynb
{
"cells": [
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"# <center> Formalizing data science via selective inference$\\:$:\n",
"# <center> Valid inference after multiple queries on the data\n",
"\n",
"### <center> Jelena Markovic, Stanford University\n",
"\n",
"\n",
"- Based on:\n",
" - Selective sampler: (Tian et al., 2016) $\\qquad$\n",
" https://arxiv.org/abs/1609.05609\n",
" - Selective bootstrap: (Markovic and Taylor, 2016) \n",
" https://arxiv.org/abs/1612.07811\n",
" - Adaptive p-values after cross-validation: (Markovic et al., 2017) \n",
" https://arxiv.org/abs/1703.06559\n",
" - Inferactive data analysis: (Bi et al., 2017) \n",
" https://arxiv.org/abs/1707.06692\n",
" "
]
},
{
"cell_type": "markdown",
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## Classical statistics\n",
"\n",
"- [Diaconis (1983)](https://statistics.stanford.edu/sites/default/files/EFS%20NSF%20206.pdf): \n",
"\n",
"> Classical mathematical statistics offers a different description of what we do (or should do) when we examine data. ... In its most rigid formulation, **we decide upon models and hypotheses before seeing the data**. Then, we compute estimates and carry out tests of our assumptions. ... However, as a description of what a real scientist does when confronting a real, rich data base, it seems as far off as a primitive ritual."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"## Practice\n",
"\n",
"- [Diaconis (1983)](https://statistics.stanford.edu/sites/default/files/EFS%20NSF%20206.pdf): \n",
"\n",
"> In practice, of course, **hypotheses are often formed after the data has been examined**; patters seen in the data combine with subject-matter knowledge in a mix that has so far defied description."
]
},
{
"cell_type": "markdown",
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## Example 1: Inference after cross-validated LASSO\n",
"- n=500, p=100, null signal"
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"from __future__ import division\n",
"import pandas as pd\n",
"import numpy as np\n",
"import matplotlib.pyplot as plt\n",
"from scipy.stats import probplot, uniform\n",
"import statsmodels.api as sm\n",
"\n",
"def plot_naive(multiple_results, label=None, fig=None):\n",
"\n",
" if fig is None:\n",
" fig = plt.figure()\n",
" ax = fig.gca()\n",
" fig.suptitle('Selective and naive pivots')\n",
"\n",
" if 'naive_pvalues' in multiple_results.columns:\n",
" ecdf_naive = sm.distributions.ECDF(multiple_results['naive_pvalues'])\n",
" G = np.linspace(0, 1)\n",
" F_naive = ecdf_naive(G)\n",
" ax.plot(G, F_naive, '-o', c='r', lw=2, label=\"Naive p-values\")\n",
" ax.plot([0, 1], [0, 1], 'k-', lw=2)\n",
" \n",
" ax.set_xlim([0, 1])\n",
" ax.set_ylim([0, 1])\n",
" ax.set_xlabel(\"Observed value\", fontsize=18)\n",
" ax.set_ylabel(\"Empirical CDF\", fontsize=18)\n",
" ax.legend(loc='lower right', fontsize=18)\n",
" fig.suptitle(\"Naive p-values\", fontsize=20)\n",
" \n",
" return fig\n",
"\n",
"results = pd.read_pickle(\"lee_et_al_pivots.pkl\")\n",
"fig = plot_naive(results)\n",
"fig.savefig('pvalue_naive.pdf')"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"- Type I error of naive (classical) p-values: 57% (target 5%)\n",
"![image](http://i.imgur.com/jq0OneC.png)"
]
},
{
"cell_type": "code",
"execution_count": 8,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"## the results from: CV_and_TG\n",
"## https://github.com/jelena-markovic/Python-software/commit/cb9e64823de42b46169cb188bf6d1b2596882b8d\n",
"\n",
"from __future__ import division\n",
"import pandas as pd\n",
"import numpy as np\n",
"import matplotlib.pyplot as plt\n",
"from scipy.stats import probplot, uniform\n",
"import statsmodels.api as sm\n",
"\n",
"def plot_all(multiple_results, multiple_results1, label=None, fig=None):\n",
"\n",
" if fig is None:\n",
" fig = plt.figure()\n",
" ax = fig.gca()\n",
" fig.suptitle('Selective and naive pivots')\n",
" \n",
" if 'pvalue' in multiple_results.columns:\n",
" ecdf = sm.distributions.ECDF(multiple_results['pvalue'])\n",
" G = np.linspace(0, 1)\n",
" F = ecdf(G)\n",
" #print(color)\n",
" ax.plot(G, F, '-o', c='b', lw=2, label=\"Lee et al. p-values\")\n",
" ax.plot([0, 1], [0, 1], 'k-', lw=2)\n",
"\n",
" if 'naive_pvalues' in multiple_results.columns:\n",
" ecdf_naive = sm.distributions.ECDF(multiple_results['naive_pvalues'])\n",
" F_naive = ecdf_naive(G)\n",
" ax.plot(G, F_naive, '-o', c='r', lw=2, label=\"Naive p-values\")\n",
" ax.plot([0, 1], [0, 1], 'k-', lw=2)\n",
" \n",
" if 'pvalue' in multiple_results1.columns:\n",
" ecdf1 = sm.distributions.ECDF(multiple_results1['pvalue'])\n",
" F1 = ecdf1(G)\n",
" #print(color)\n",
" ax.plot(G, F1, '-o', c='g', lw=2, label=\"CV adjusted p-values\")\n",
" ax.plot([0, 1], [0, 1], 'k-', lw=2)\n",
"\n",
" \n",
" ax.set_xlim([0, 1])\n",
" ax.set_ylim([0, 1])\n",
" ax.set_xlabel(\"Observed value\", fontsize=18)\n",
" ax.set_ylabel(\"Empirical CDF\", fontsize=18)\n",
" ax.legend(loc='lower right', fontsize=18)\n",
" fig.suptitle(\"Selective and naive p-values\", fontsize=20)\n",
" \n",
" return fig\n",
"\n",
"results = pd.read_pickle(\"lee_et_al_pivots.pkl\")\n",
"results1 = pd.read_pickle(\"cv_corrected_nonrandomized_lasso.pkl\")\n",
"fig = plot_all(results, results1)\n",
"fig.savefig('pvalue_all.pdf')"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## Classical statistics\n",
"\n",
"- No guarantees if the same data is used for selection and inference.\n",
"- P-values and confidence intervals no longer valid."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"# Selective inference\n",
"\n",
"- Valid post-selection inference based on the same data you used for selection.\n",
"\n",
"- We need to account for only reporting inference for the selected coefficients.\n",
"\n",
"- Base inference using the conditional distribution of the data.\n",
"\n",
"- Conditioning is on the outcomes of the model selection procedure, everything you **look at** at the model selection stage.\n",
"\n"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## Outline:\n",
"\n",
" - *Example 1*: Cross-validated LASSO\n",
" - Selective sampler\n",
" - Selective inference framework\n",
" - *Example 2.0*: Group LASSO\n",
" - *Example 2.1*: Adding a fresh dataset\n",
" - *Example 3*: Multiple queries (marginal screening + LASSO)"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## Example 1: Inference after cross-validated LASSO"
]
},
{
"cell_type": "markdown",
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"### Step 1: Adjust for LASSO (treat $\\lambda$ as fixed)\n",
"\n",
"- Truncated Gaussian test statistic valid after selection (Lee et al., 2016)"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"- Type I error rate: Lee et al. p-values 16% (target 5%)\n",
"\n",
"<img src=\"http://i.imgur.com/rcEeMpj.png\" width=\"500\">\n"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"### Step 2: Adjust for CV as well \n",
" - By conditioning on the minimizer of the CV error curve (Markovic et al., 2017)"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"![image](http://i.imgur.com/ZMcG9qg.png)"
]
},
{
"cell_type": "code",
"execution_count": 71,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAZAAAAEpCAYAAAC9enRxAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
"text/plain": [
"<matplotlib.figure.Figure at 0x11e31b7d0>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"fig = plot_all(results, results1)"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## Power comparison\n",
"\n",
"- Setup of Barber and Candes (2015): n=3000, p=1000, s=30 (non-zero coefficients 3.5) \n",
"\n",
"\n",
"\n",
"| | <font size=\"5\"> FDR </font>| <font size=\"5\"> Power </font> | \n",
"| ------------- |-------|:-------------:|\n",
"| <font size=\"5\"> Knockoffs</font> | <font size=\"5\"> 0.183</font> | <font size=\"5\">0.654 </font>|\n",
"| <font size=\"5\">Randomized LASSO + CV + BHq </font> |<font size=\"5\"> 0.208 </font> |<font size=\"5\"> 0.606 </font>|\n",
"\n",
"<!--\n",
"| <font size=\"5\">LASSO + CV + BHq </font> |<font size=\"5\"> 0.068 </font> | <font size=\"5\">0.135 </font> | \n",
"-->"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## Inference after randomized LASSO via selective sampler"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"### Randomized LASSO\n",
"- Randomized LASSO objective: given $((X,y),\\omega)\\sim F\\times G$\n",
"$$\\begin{pmatrix} \\widehat{\\beta}_E \\\\ 0 \\end{pmatrix}=\\widehat{\\beta}=\n",
"\\underset{\\beta\\in\\mathbb{R}^p}{\\text{argmin}}\\frac{1}{2}\\|y-X\\beta\\|_2^2+\\lambda\\|\\beta\\|_1-\\omega^T\\beta+\\frac{\\epsilon}{2}\\|\\beta\\|_2^2.$$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"#### Randomization reconstruction:\n",
"\n",
"- Denote: $D=\\begin{pmatrix} \\bar{\\beta}_E \\\\ X_{-E}^T(y-X_E\\bar{\\beta}_E) \\end{pmatrix},\\;\\; \\bar{\\beta}_E=(X_E^TX_E)^{-1}X_E^Ty$.\n",
"\n",
"- KKT conditions: \n",
"\n",
" $\\omega = -X^T(y-X_E\\beta_E)+\\lambda\\begin{pmatrix} s_E \\\\ u_{-E}\\end{pmatrix}+\\epsilon\\begin{pmatrix} \\widehat{\\beta}_E \\\\ 0 \\end{pmatrix} \\\\ $\n",
" $=-\\begin{pmatrix} X_E^TX_E & 0 \\\\ X_{-E}^TX_E & I_{p-|E|} \\end{pmatrix}D+ \\begin{pmatrix}X_E^TX_E+\\epsilon I_{|E|} \\\\ X_{-E}^TX_E \\end{pmatrix}\\widehat{\\beta}_E+\\lambda\\begin{pmatrix} s_E \\\\ u_{-E} \\end{pmatrix}$ \n",
" $ =\\omega(D,\\widehat{\\beta}_E, u_{-E}) $ \n",
" \n",
" \n",
" with the constraints:\n",
" $\\text{sign}(\\widehat{\\beta}_E)=s_E, \\|u_{-E}\\|_{\\infty}\\leq 1$.\n",
" "
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## Selective sampler\n",
"- Change of probabily density.\n",
"- <font color='blue'>Original space</font>: (data=$D$, randomization=$\\omega$) conditional on selection region $LASSO(D,\\omega)=E$.\n",
"- <font color='blue'>Simpler space</font>: (data=$D$, optimization variables=$(\\widehat{\\beta}_E,u_{-E})$) with the constraints on the optimization variables."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"### Selective density\n",
"\n",
"- The selective density of $(D,\\omega)$ is proportional to\n",
"\n",
"$$\\phi_{(\\mu_D,\\Sigma_D)}(D)\\cdot g(\\omega)\\cdot\\mathbb{I}_{\\{LASSO(D,\\omega)=E\\}}.$$\n",
"\n",
"- The randomization reconstruction coming from KKT \n",
"\n",
"$$\\omega=\\omega(D,\\widehat{\\beta}_E,u_{-E}).$$\n",
"\n",
"- Selective density of $(D, \\widehat{\\beta}_E, u_{-E})$ is proportional to\n",
"\n",
"$$\\phi_{(\\mu_D,\\Sigma_D)}(D)\\cdot g(\\omega(D,\\widehat{\\beta}_E,u_{-E}))\\cdot\\mathbb{I}_{\\{\\text{sign}(\\widehat{\\beta}_E)=s_E\\}}\\cdot\\mathbb{I}_{\\{\\|u_{-E}\\|_{\\infty}\\leq 1\\}}.$$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## Inference\n",
"- Target parameter $\\beta_E^*=(\\mathbb{E}_F[X_E^TX_E])^{-1}\\mathbb{E}_F[X_E^Ty]$."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"- Pre-selection CLT for $T=\\bar{\\beta}_E$ (OLS $y\\sim X_E$): \n",
"\n",
"$$ T-\\beta_E^*\\rightarrow\\mathcal{N}(0,\\Sigma_T).$$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"- Post-selection inference using the conditional distribution of $T$."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"- Linear decomposition of $D$ in terms of $T$:\n",
"\n",
" $$D=N_D+\\Sigma_{D,T}\\Sigma_T^{-1}T.$$"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"- Sample $(T,\\widehat{\\beta}_E, u_{-E})$ from the selective density proportional to\n",
"\n",
"$$\\phi_{(\\beta_E^*,\\Sigma_T)}(T)\\cdot g(\\omega(N_D+\\Sigma_{D,T}\\Sigma_T^{-1}T, \\widehat{\\beta}_E, u_{-E}))$$\n",
"\n",
"$\\qquad$ with the constraints on $\\widehat{\\beta}_E$ and $u_{-E}$."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## Selective inference framework\n",
"\n",
"$$\\;$$\n",
"$$\\;$$\n",
"\\begin{equation*}\n",
"\\begin{array}{|c|c|c|}\n",
" & \\text{General} & \\text{LASSO} \\\\ \\hline\n",
" \\text{Dataset} & S\\sim F & (X,y)\\sim F \\\\ \\hline\n",
" \\text{Selection alg.} & \\widehat{M}(S,\\omega)=M & LASSO(D,\\omega)=E \\\\ \\hline\n",
" \\text{Selection region} & \\begin{matrix}\\{(S',\\omega'): \\\\ \\widehat{M}(S',\\omega')=M\\} \\end{matrix} & \\begin{matrix} \\{(D',\\omega'):\\\\ LASSO(D',\\omega')=E\\} \\end{matrix} \\\\ \\hline\n",
" \\text{Target param.} & \\theta=\\theta(F,E) & \\beta_E^* \\\\\n",
" \\text{Target stat.} & T & \\bar{\\beta}_E \\\\ \n",
" \\text{Pre-selection} & T-\\theta\\rightarrow\\mathcal{N}(0,\\Sigma_T) & \\bar{\\beta}_E-\\beta_E^*\\rightarrow\\mathcal{N}(0,\\Sigma_T) \\\\ \n",
" \\text{Post-selection} & \\mathcal{L}(T|\\widehat{M}(S)=M) & \\mathcal{L}(\\bar{\\beta}_E|E) \\\\\n",
"\\end{array}\n",
"\\end{equation*}"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## Example 2: Infererence after randomized Group LASSO\n",
"- n=600, p=100, 10 groups total of 10 variables each"
]
},
{
"cell_type": "code",
"execution_count": 96,
"metadata": {
"collapsed": true,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"# master branch https://github.com/jelena-markovic/Python-software/commits/master\n",
"import numpy as np\n",
"import regreg.api as rr \n",
"from selection.tests.instance import gaussian_instance\n",
"from selection.api import (randomization, \n",
" glm_group_lasso, \n",
" multiple_queries, \n",
" glm_target)\n",
"from selection.randomized.M_estimator import restricted_Mest\n",
"from selection.randomized.query import (naive_pvalues, naive_confidence_intervals)"
]
},
{
"cell_type": "code",
"execution_count": 143,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"np.random.seed(18)\n",
"n=600 \n",
"p=100 \n",
"s=0\n",
"rho=0. \n",
"signal=1\n",
"lam_frac=8.\n",
"X, y, beta, nonzero, sigma = gaussian_instance(n=n, p=p, s=s, rho=rho, signal=signal, sigma=1)\n",
"lam = lam_frac*np.mean(np.fabs(np.dot(X.T, np.random.standard_normal((n, 1000))))) * sigma\n",
"nonzero = np.where(beta)[0]\n",
"loss = rr.glm.gaussian(X, y)\n",
"epsilon = 1./np.sqrt(n)\n",
"W = np.ones(p)*lam"
]
},
{
"cell_type": "code",
"execution_count": 144,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "fragment"
}
},
"outputs": [],
"source": [
"groups = np.concatenate([np.arange(10) for i in range(p/10)])\n",
"penalty = rr.group_lasso(groups, weights=dict(zip(np.arange(p), W)), lagrange=1.)\n",
"randomizer = randomization.isotropic_gaussian((p,), scale=1.)\n",
"M_est1 = glm_group_lasso(loss, epsilon, penalty, randomizer)\n",
"mv = multiple_queries([M_est1])"
]
},
{
"cell_type": "code",
"execution_count": 145,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "fragment"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"('active set', array([ 6, 16, 26, 36, 46, 56, 66, 76, 86, 96]))\n"
]
}
],
"source": [
"mv.solve()\n",
"active_union = M_est1.selection_variable['variables']\n",
"print(\"active set\", np.nonzero(active_union)[0])"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"### Setting up the sampler"
]
},
{
"cell_type": "code",
"execution_count": 146,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "fragment"
}
},
"outputs": [],
"source": [
"target_sampler, target_observed = glm_target(loss,\n",
" active_union,\n",
" mv,\n",
" bootstrap=False)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"collapsed": true,
"slideshow": {
"slide_type": "fragment"
}
},
"outputs": [],
"source": [
"target_sample = target_sampler.sample(ndraw=10000, burnin=2000)"
]
},
{
"cell_type": "code",
"execution_count": 147,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"[[-1.04510451 3.63536354]\n",
" [-2.89528953 2.37523752]\n",
" [-2.86528653 2.17521752]\n",
" [-1.94519452 3.08530853]\n",
" [-2.85528553 1.62516252]\n",
" [-5.06550655 -0.3850385 ]\n",
" [-0.77507751 3.73537354]\n",
" [-3.54535454 1.83518352]\n",
" [-3.58535854 1.18511851]\n",
" [-1.01510151 3.69536954]]\n",
"[ 0.36399152 0.80653521 0.82771933 0.65145272 0.63967232 0.0551894\n",
" 0.28625287 0.59622237 0.40426566 0.36151533]\n"
]
}
],
"source": [
"active_set = np.nonzero(active_union)[0]\n",
"true_vec = beta[active_union]\n",
"LU = target_sampler.confidence_intervals(target_observed,\n",
" sample=target_sample,\n",
" level=0.9)\n",
"pvalues = target_sampler.coefficient_pvalues(target_observed,\n",
" parameter=np.zeros_like(true_vec),\n",
" sample=target_sample)\n",
"LU_naive = naive_confidence_intervals(target_sampler, target_observed)\n",
"naive_pvals = naive_pvalues(target_sampler, target_observed, true_vec)\n",
"print(LU)\n",
"print(pvalues)"
]
},
{
"cell_type": "code",
"execution_count": 117,
"metadata": {
"collapsed": true,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"import pandas\n",
"import itable\n",
"def print_table(LU1, LU2, label1, label2, active_union):\n",
" active_set = np.nonzero(active_union)[0]\n",
" LU_1=['('+str(np.round(LU1[i,0],2))+','+str(round(LU1[i,1],2))+')' for i in range(LU1.shape[0])]\n",
" LU_2=['('+str(np.round(LU2[i,0],2))+','+str(round(LU2[i,1],2))+')' for i in range(LU2.shape[0])]\n",
" CI = pandas.DataFrame({'Active':active_set, label1:LU_1, label2:LU_2})\n",
" CI_table = itable.PrettyTable(CI, tstyle=itable.TableStyle(theme=\"theme1\"), center=True)\n",
" non_signif_split = [LU1[i,0] < 0 and LU1[i,1]>0 for i in range(LU1.shape[0])]\n",
" non_signif_carving = [LU2[i,0] < 0 and LU2[i,1]>0 for i in range(LU2.shape[0])]\n",
" non_signif_split = np.asarray(non_signif_split)\n",
" non_signif_carving = np.asarray(non_signif_carving)\n",
" CI_table.set_cell_style(cols=[2], rows=np.nonzero(~non_signif_split)[0], color='red', font_weight='bold')#, format_function=lambda p: \"%0.3f\" % p)\n",
" #CI_table.set_cell_style(cols=[2], rows=np.nonzero(~signif_split)[0], color='blue') #, format_function=lambda p: \"%0.3f\" % p)\n",
" CI_table.set_cell_style(cols=[1], rows=np.nonzero(~non_signif_carving)[0], color='red', font_weight='bold')#, format_function=lambda p: \"%0.3f\" % p)\n",
" #CI_table.set_cell_style(cols=[1], rows=np.nonzero(~signif_carving)[0], color='blue') #, format_function=lambda p: \"%0.3f\" % p)\n",
" CI_table.padding_width=0.5\n",
" return CI_table"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"### Selective and naive confidence intervals"
]
},
{
"cell_type": "code",
"execution_count": 148,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "fragment"
}
},
"outputs": [
{
"data": {
"text/html": [
"<center><table style=\"color: black;border: 1px solid black;\"><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;font-weight: bold;background-color: lightgray;\">Active</td><td style=\"color: black;font-weight: bold;background-color: lightgray;\">naive</td><td style=\"color: black;font-weight: bold;background-color: lightgray;\">selective</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">6</td><td style=\"color: black;border: 1px solid black;\">(-1.46,1.93)</td><td style=\"color: black;border: 1px solid black;\">(-1.05,3.64)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">16</td><td style=\"color: black;border: 1px solid black;\">(-3.35,0.22)</td><td style=\"color: black;border: 1px solid black;\">(-2.9,2.38)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">26</td><td style=\"color: red;font-weight: bold;\">(-3.77,-0.11)</td><td style=\"color: black;border: 1px solid black;\">(-2.87,2.18)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">36</td><td style=\"color: red;font-weight: bold;\">(0.45,3.9)</td><td style=\"color: black;border: 1px solid black;\">(-1.95,3.09)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">46</td><td style=\"color: black;border: 1px solid black;\">(-0.91,2.46)</td><td style=\"color: black;border: 1px solid black;\">(-2.86,1.63)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">56</td><td style=\"color: red;font-weight: bold;\">(-3.81,-0.41)</td><td style=\"color: red;font-weight: bold;\">(-5.07,-0.39)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">66</td><td style=\"color: black;border: 1px solid black;\">(-0.92,2.37)</td><td style=\"color: black;border: 1px solid black;\">(-0.78,3.74)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">76</td><td style=\"color: black;border: 1px solid black;\">(-2.0,1.69)</td><td style=\"color: black;border: 1px solid black;\">(-3.55,1.84)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">86</td><td style=\"color: black;border: 1px solid black;\">(-2.44,1.05)</td><td style=\"color: black;border: 1px solid black;\">(-3.59,1.19)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">96</td><td style=\"color: black;border: 1px solid black;\">(-2.28,1.07)</td><td style=\"color: black;border: 1px solid black;\">(-1.02,3.7)</td></tr></table></center>"
],
"text/plain": [
"<itable.itable.PrettyTable at 0x11853b4d0>"
]
},
"execution_count": 148,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"print_table(LU, LU_naive, \"selective\", \"naive\", active_union)"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## Group LASSO p-values"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"\n",
"| |<font size=\"4\"> Type I error (target 5%)</font>| <font size=\"4\"> Coverage (target 90%) </font>| <font size=\"4\"> Average CI length </font>| \n",
"| ------------- |:-------------:|:-------------:|:-------------:|\n",
"|<font size=\"4\"> Naive</font>|<font size=\"5\">15%</font>|<font size=\"5\"> 76%</font>|<font size=\"5\"> 3.27 </font>|\n",
"|<font size=\"4\"> Selective </font>|<font size=\"5\"> 5% </font>|<font size=\"5\">90% </font> |<font size=\"5\">4.42 </font> |\n",
"\n",
"![image](http://i.imgur.com/7PE8xl2.png)"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## Adding a fresh dataset\n",
"\n",
"\n",
"| |<font size=\"4\"> Type I error (target 5%)</font>|<font size=\"4\"> Coverage (target 90%)</font>|<font size=\"4\"> Average CI length </font>| \n",
"| ------------- |:-------------:|:-------------:|:-------------:|\n",
"|<font size=\"4\">Data splitting </font>|<font size=\"5\">5% </font>|<font size=\"5\">88% </font>|<font size=\"5\">3.27</font>|\n",
"|<font size=\"4\">Selective</font>|<font size=\"5\">5%</font>|<font size=\"5\">88% </font>|<font size=\"5\">2.84</font>|\n",
"\n",
"![image](http://i.imgur.com/PXgg5QX.png)"
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"import pandas as pd\n",
"from selection.tests import reports\n",
"runs_read = pd.read_pickle(\"test_new_data.pkl\")\n",
"fig = reports.split_pvalue_plot(runs_read)\n",
"fig.suptitle('Inference after adding fresh data', fontsize=20)\n",
"fig.savefig(\"test_new_data.pdf\")"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"# Example 3: HIV dataset\n",
"\n",
"- In vitro measurement of resistance of sample of HIV viruses to NRTI drug 3TC.\n",
"\n",
"- 633 cases, and 91 different mutations occuring more than 10 times in sample.\n",
"\n",
"- Source: [HIVDB](http://hivdb.stanford.edu/pages/published_analysis/genophenoPNAS2006/)\n",
"\n",
"- Goal: to build an <font class=\"emphblue\">interpretable predictive model</font> of resistance based on mutation pattern."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"### Data set\n",
"\n",
"- Design matrix X: n=633, p=91\n",
"- response y: (site / amino acid) pairs"
]
},
{
"cell_type": "code",
"execution_count": 80,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"%%capture\n",
"from IPython.display import HTML\n",
"import os, numpy as np, pandas, statsmodels.api as sm\n",
"import regreg.api as rr\n",
"\n",
"from selection.api import (randomization, \n",
" glm_group_lasso,\n",
" split_glm_group_lasso, \n",
" multiple_queries, \n",
" glm_target)\n",
"from selection.randomized.glm import pairs_bootstrap_glm, bootstrap_cov\n",
"from selection.randomized.query import naive_confidence_intervals, naive_pvalues"
]
},
{
"cell_type": "code",
"execution_count": 81,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"np.random.seed(0)\n",
"HTML(\"\"\"\n",
"<style type=\"text/css\">\n",
" .emphred {color: #cc2222; font-size: 100%; font-weight: bold; font-family: arial,helvetica,sans-serif}\n",
" .emphblue {color: #2222cc; font-size: 100%; font-weight: bold; font-family: arial,helvetica,sans-serif}\n",
"</style>\n",
"\"\"\")\n",
"\n",
"# Run with commit 55fb577e418e8ca15eb55728e12be872aa630ab4\n",
"# from http://github.com/jonathan-taylor/selective-inference\n",
"if not os.path.exists(\"NRTI_DATA.txt\"):\n",
" NRTI = pandas.read_table(\"http://hivdb.stanford.edu/pages/published_analysis/genophenoPNAS2006/DATA/NRTI_DATA.txt\", na_values=\"NA\")\n",
"else:\n",
" NRTI = pandas.read_table(\"NRTI_DATA.txt\")\n",
"NRTI_specific = []\n",
"NRTI_muts = []\n",
"mixtures = np.zeros(NRTI.shape[0])\n",
"for i in range(1,241):\n",
" d = NRTI['P%d' % i]\n",
" for mut in np.unique(d):\n",
" if mut not in ['-','.'] and len(mut) == 1:\n",
" test = np.equal(d, mut)\n",
" if test.sum() > 10:\n",
" NRTI_specific.append(np.array(np.equal(d, mut))) \n",
" NRTI_muts.append(\"P%d%s\" % (i,mut))\n",
"\n",
"NRTI_specific = NRTI.from_records(np.array(NRTI_specific).T, columns=NRTI_muts)\n",
"\n",
"# Next, standardize the data, keeping only those where Y is not missing\n",
"\n",
"X_NRTI = np.array(NRTI_specific, np.float)\n",
"Y = NRTI['3TC'] # shorthand\n",
"keep = ~np.isnan(Y).astype(np.bool)\n",
"X_NRTI = X_NRTI[np.nonzero(keep)]; Y=Y[keep]\n",
"Y = np.array(np.log(Y), np.float); Y -= Y.mean()\n",
"X_NRTI -= X_NRTI.mean(0)[None, :]; X_NRTI /= X_NRTI.std(0)[None,:]\n",
"X = X_NRTI # shorthand\n",
"n, p = X.shape\n",
"X /= np.sqrt(n)\n",
"\n",
"ols_fit = sm.OLS(Y, X).fit()\n",
"sigma_3TC = np.linalg.norm(ols_fit.resid) / np.sqrt(n-p-1)\n",
"OLS_3TC = ols_fit.params"
]
},
{
"cell_type": "code",
"execution_count": 82,
"metadata": {
"collapsed": true,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"np.random.seed(1)\n",
"n, p = X.shape\n",
"# Choose a LASSO parameter\n",
"lambda_theoretical = []\n",
"for i in range(10000):\n",
" lambda_theoretical.append(np.fabs(np.dot(X.T, np.random.standard_normal(n))).max())\n",
"lambda_theoretical = np.round((1 * np.mean(lambda_theoretical) * sigma_3TC))\n",
"epsilon = 1. / np.sqrt(n)"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "skip"
}
},
"source": [
"## Data splitting: Run Lasso on 80% of the data"
]
},
{
"cell_type": "code",
"execution_count": 83,
"metadata": {
"collapsed": true,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"np.random.seed(1)\n",
"W = 0.6*np.ones(p)*lambda_theoretical\n",
"penalty = rr.group_lasso(np.arange(p), \n",
" weights=dict(zip(np.arange(p), W)), \n",
" lagrange=1.)"
]
},
{
"cell_type": "code",
"execution_count": 84,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"loss = rr.glm.gaussian(X, Y)\n",
"split_frac = 0.8\n",
"m = int(split_frac * n)\n",
"M_est1 = split_glm_group_lasso(loss, epsilon, m, penalty)\n",
"mv = multiple_queries([M_est1])\n",
"mv.solve()"
]
},
{
"cell_type": "code",
"execution_count": 85,
"metadata": {
"collapsed": true,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"active_union = M_est1.selection_variable['variables']\n",
"nactive = np.sum(active_union)\n",
"active_set = np.nonzero(active_union)[0]\n",
"NRTI_muts=np.array(NRTI_muts)\n",
"active_set_labels = NRTI_muts[active_union]"
]
},
{
"cell_type": "code",
"execution_count": 86,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"24\n",
"['P41L' 'P62V' 'P65R' 'P67N' 'P68G' 'P69D' 'P69i' 'P75I' 'P75T' 'P77L'\n",
" 'P83K' 'P90I' 'P115F' 'P118I' 'P162C' 'P174H' 'P181C' 'P184V' 'P210W'\n",
" 'P211K' 'P215F' 'P215Y' 'P219R' 'P228H']\n"
]
}
],
"source": [
"print(len(active_set))\n",
"print(active_set_labels)"
]
},
{
"cell_type": "code",
"execution_count": 87,
"metadata": {
"collapsed": true,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"def split_inference(glm_loss, X, y , active, leftout_indices, alpha=0.1):\n",
"\n",
" import regreg.api as rr\n",
"\n",
" loss = glm_loss(X[leftout_indices, ], y[leftout_indices])\n",
" boot_target, target_observed = pairs_bootstrap_glm(loss, active)\n",
" nactive = np.sum(active)\n",
" size= np.sum(leftout_indices)\n",
" observed = target_observed[:nactive]\n",
" boot_target_restricted = lambda indices: boot_target(indices)[:nactive]\n",
" sampler = lambda: np.random.choice(size, size=(size,), replace=True)\n",
" target_cov = bootstrap_cov(sampler, boot_target_restricted)\n",
"\n",
" from scipy.stats import norm as ndist\n",
" quantile = - ndist.ppf(alpha / float(2))\n",
" LU = np.zeros((2, nactive))\n",
" pvalues = np.zeros(nactive)\n",
" for j in range(observed.shape[0]):\n",
" sigma = np.sqrt(target_cov[j, j])\n",
" LU[0, j] = observed[j] - sigma * quantile\n",
" LU[1, j] = observed[j] + sigma * quantile\n",
" pval = ndist.cdf(observed[j]/sigma)\n",
" pvalues[j] = 2 * min(pval, 1-pval)\n",
" return target_observed, LU.T, pvalues\n",
"\n",
"def average_length(LU):\n",
" return sum(LU[:,1]-LU[:,0])/float(LU.shape[0])"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "skip"
}
},
"source": [
"## Data splitting: Inference on 20% of the data"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"collapsed": true,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"np.random.seed(1)"
]
},
{
"cell_type": "code",
"execution_count": 88,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"average length 6.83003533566\n"
]
}
],
"source": [
"leftout_indices = M_est1.randomized_loss.saturated_loss.case_weights == 0\n",
"split_observed, LU_split, pvalues_split = split_inference(rr.glm.gaussian, X, Y, active_union, leftout_indices)\n",
"print \"average length\", average_length(LU_split)"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "skip"
}
},
"source": [
"## Data splitting confidence intervals and p-values"
]
},
{
"cell_type": "code",
"execution_count": 113,
"metadata": {
"collapsed": true,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"# plot CIs\n",
"import matplotlib.pyplot as plt\n",
"%matplotlib inline\n",
"\n",
"def plot_CI(labels, target_observed1, CI1, pvalues1, plot_title, target_observed2=None, CI2=None, pvalues2=None):\n",
" nactive = target_observed1.shape[0]\n",
" fig, ax = plt.subplots(figsize=(20, 10))\n",
" ax.set_ylabel('Parameter values', fontsize=18)\n",
" ax.set_title('Confidence intervals and observed values', fontsize=20)\n",
" ymin = -2\n",
" ymax = 15\n",
" ax.set_ylim((ymin,ymax))\n",
" \n",
" errors1 = [target_observed1 - CI1[:, 0], CI1[:, 1] - target_observed1]\n",
" ind1 = np.arange(nactive) # the x locations for the groups\n",
" width1 = 0.2 # the width of the bars\n",
" rects1 = ax.bar(ind1, target_observed1, width1, color='r', yerr=errors1, label = plot_title)\n",
" \n",
" # add axes ticks\n",
" ax.set_xticks(ind1 + (width1/2))\n",
" ax.set_xticklabels(tuple(labels), rotation = 90, fontsize=15)\n",
"\n",
" def autolabel(rects, pvalues):\n",
" # attach some text labels\n",
" for i in range(len(rects)):\n",
" height = min(rects[i].get_height(), ymax)\n",
" ax.text(rects[i].get_x() + rects[i].get_width()/2., height + 1.2,\n",
" \"{:4.2f}\".format(pvalues[i]),\n",
" ha='center', va='bottom', rotation=90, fontsize = 12)\n",
" \n",
" autolabel(rects1, pvalues1)\n",
" \n",
" if target_observed2 is not None:\n",
" errors2 = [target_observed2 - CI2[:, 0], CI2[:, 1] - target_observed2]\n",
" ind2 = np.arange(nactive)+0.5\n",
" width2 = 0.2\n",
" rects2 = ax.bar(ind2, target_observed2, width2, color='y', yerr=errors2, label = \"Data carving\")\n",
" autolabel(rects2, pvalues2)\n",
" \n",
" ax.legend(loc='upper right', fontsize=18)\n",
" plt.savefig(''.join([plot_title, \".pdf\"]))\n",
" #plt.show()"
]
},
{
"cell_type": "code",
"execution_count": 114,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAABJQAAAK2CAYAAAAYHsZCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nrvx/tt040adBFBgAAAAASUVORK5CYII=\n",
"text/plain": [
"<matplotlib.figure.Figure at 0x1176457d0>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"plot_CI(active_set_labels, split_observed, LU_split, pvalues_split, \"Data splitting\")"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "skip"
}
},
"source": [
"## Data carving: Inference using the whole data\n",
"#### Conditional on the selection"
]
},
{
"cell_type": "code",
"execution_count": 93,
"metadata": {
"collapsed": true,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"np.random.seed(1)"
]
},
{
"cell_type": "code",
"execution_count": 94,
"metadata": {
"collapsed": true,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"target_sampler_boot, target_observed = glm_target(loss,\n",
" active_union,\n",
" mv,\n",
" bootstrap=True)\n",
"target_sample_boot = target_sampler_boot.sample(ndraw=20000, burnin=0)\n"
]
},
{
"cell_type": "code",
"execution_count": 95,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"('average length', 4.873714227405606)\n"
]
}
],
"source": [
"LU_boot = target_sampler_boot.confidence_intervals(target_observed, \n",
" sample=target_sample_boot)\n",
"\n",
"p_boot = target_sampler_boot.coefficient_pvalues(target_observed,\n",
" parameter=np.zeros(nactive),\n",
" sample=target_sample_boot)\n",
"\n",
"print(\"average length\", average_length(LU_boot))"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "skip"
}
},
"source": [
"## Data carving confidence intervals and p-values"
]
},
{
"cell_type": "code",
"execution_count": 96,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"## saving the results for the plot\n",
"import pandas as pd\n",
"dfs = []\n",
"active_var = [i for i in range(active_union.shape[0]) if active_union[i]==True]\n",
"dfs.append(np.array(active_var))\n",
"dfs.append(target_observed)\n",
"dfs.append(LU_boot)\n",
"dfs.append(p_boot)\n",
"dfs.append(active_set_labels)\n",
"dfs = pd.DataFrame(dfs)\n",
"dfs.to_pickle(\"Data_carving.pkl\")\n"
]
},
{
"cell_type": "code",
"execution_count": 115,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"## read the results\n",
"import pandas as pd\n",
"results = pd.read_pickle(\"Data_carving.pkl\")\n",
"active_var= list(results.ix[0])\n",
"nactive = len(active_var)\n",
"CI = np.zeros((nactive,2))\n",
"CI[:,0]= np.array([list(results.ix[2])[i][0] for i in range(nactive)])\n",
"CI[:,1] = np.array([list(results.ix[2])[i][1] for i in range(nactive)])\n",
"active_set_labels = list(results.ix[4])"
]
},
{
"cell_type": "code",
"execution_count": 116,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAABJQAAAK2CAYAAAAYHsZCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nERERERnI/wfm4dKmmJG6YgAAAABJRU5ErkJggg==\n",
"text/plain": [
"<matplotlib.figure.Figure at 0x11e728f10>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"plot_CI(active_set_labels, np.array(list(results.ix[1])), CI, list(results.ix[3]), \"Data carving\")"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "skip"
}
},
"source": [
"## Data carving vs. data splitting \n",
"### Confidence Intervals"
]
},
{
"cell_type": "code",
"execution_count": 23,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"import itable\n",
"active_set = np.nonzero(active_union)[0]\n",
"LU_B=['('+str(np.round(LU_boot[i,0],2))+','+str(round(LU_boot[i,1],2))+')' for i in range(LU_boot.shape[0])]\n",
"LU_S=['('+str(np.round(LU_split[i,0],2))+','+str(round(LU_split[i,1],2))+')' for i in range(LU_split.shape[0])]\n",
"CI = pandas.DataFrame({'Active Mutations':active_set, 'Data carving':LU_B, 'Data splitting':LU_S})\n",
"CI_table = itable.PrettyTable(CI, tstyle=itable.TableStyle(theme=\"theme1\"), center=True)\n",
"non_signif_split = [LU_split[i,0] < 0 and LU_split[i,1]>0 for i in range(LU_split.shape[0])]\n",
"non_signif_carving = [LU_boot[i,0] < 0 and LU_boot[i,1]>0 for i in range(LU_boot.shape[0])]\n",
"non_signif_split = np.asarray(non_signif_split)\n",
"non_signif_carving = np.asarray(non_signif_carving)\n",
"CI_table.set_cell_style(cols=[2], rows=np.nonzero(~non_signif_split)[0], color='red', font_weight='bold')#, format_function=lambda p: \"%0.3f\" % p)\n",
"#CI_table.set_cell_style(cols=[2], rows=np.nonzero(~signif_split)[0], color='blue') #, format_function=lambda p: \"%0.3f\" % p)\n",
"CI_table.set_cell_style(cols=[1], rows=np.nonzero(~non_signif_carving)[0], color='red', font_weight='bold')#, format_function=lambda p: \"%0.3f\" % p)\n",
"#CI_table.set_cell_style(cols=[1], rows=np.nonzero(~signif_carving)[0], color='blue') #, format_function=lambda p: \"%0.3f\" % p)\n",
"CI_table.padding_width=0.5"
]
},
{
"cell_type": "code",
"execution_count": 24,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [
{
"data": {
"text/html": [
"<center><table style=\"color: black;border: 1px solid black;\"><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;font-weight: bold;background-color: lightgray;\">Active Mutations</td><td style=\"color: black;font-weight: bold;background-color: lightgray;\">Data carving</td><td style=\"color: black;font-weight: bold;background-color: lightgray;\">Data splitting</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">7</td><td style=\"color: red;font-weight: bold;\">(-7.11,-0.67)</td><td style=\"color: black;border: 1px solid black;\">(-7.92,0.21)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">15</td><td style=\"color: black;border: 1px solid black;\">(-3.4,3.4)</td><td style=\"color: black;border: 1px solid black;\">(-5.47,1.61)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">16</td><td style=\"color: red;font-weight: bold;\">(6.22,9.39)</td><td style=\"color: red;font-weight: bold;\">(5.68,11.13)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">18</td><td style=\"color: red;font-weight: bold;\">(3.24,6.88)</td><td style=\"color: red;font-weight: bold;\">(3.41,8.21)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">19</td><td style=\"color: black;border: 1px solid black;\">(-5.02,0.99)</td><td style=\"color: black;border: 1px solid black;\">(-4.5,1.4)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">20</td><td style=\"color: black;border: 1px solid black;\">(-2.98,1.4)</td><td style=\"color: black;border: 1px solid black;\">(-2.83,1.42)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">22</td><td style=\"color: red;font-weight: bold;\">(3.29,7.62)</td><td style=\"color: red;font-weight: bold;\">(1.88,10.44)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">26</td><td style=\"color: black;border: 1px solid black;\">(-1.2,2.93)</td><td style=\"color: black;border: 1px solid black;\">(-0.06,3.06)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">28</td><td style=\"color: red;font-weight: bold;\">(1.1,6.56)</td><td style=\"color: red;font-weight: bold;\">(0.73,6.59)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">29</td><td style=\"color: red;font-weight: bold;\">(0.5,3.84)</td><td style=\"color: black;border: 1px solid black;\">(-0.63,3.37)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">30</td><td style=\"color: red;font-weight: bold;\">(-2.87,-1.13)</td><td style=\"color: black;border: 1px solid black;\">(-2.48,1.75)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">31</td><td style=\"color: red;font-weight: bold;\">(1.18,6.15)</td><td style=\"color: red;font-weight: bold;\">(1.19,7.17)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">40</td><td style=\"color: red;font-weight: bold;\">(0.08,3.21)</td><td style=\"color: black;border: 1px solid black;\">(-0.73,5.16)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">42</td><td style=\"color: black;border: 1px solid black;\">(-1.46,1.76)</td><td style=\"color: black;border: 1px solid black;\">(-2.45,2.86)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">56</td><td style=\"color: black;border: 1px solid black;\">(-6.9,1.91)</td><td style=\"color: black;border: 1px solid black;\">(-0.3,3.48)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">60</td><td style=\"color: black;border: 1px solid black;\">(-0.28,4.52)</td><td style=\"color: black;border: 1px solid black;\">(-17.28,15.35)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">66</td><td style=\"color: black;border: 1px solid black;\">(-0.14,5.33)</td><td style=\"color: red;font-weight: bold;\">(0.88,8.88)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">67</td><td style=\"color: red;font-weight: bold;\">(54.82,56.87)</td><td style=\"color: red;font-weight: bold;\">(53.96,58.5)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">77</td><td style=\"color: black;border: 1px solid black;\">(-3.59,0.72)</td><td style=\"color: black;border: 1px solid black;\">(-3.9,1.7)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">78</td><td style=\"color: black;border: 1px solid black;\">(-3.33,2.13)</td><td style=\"color: black;border: 1px solid black;\">(-1.61,3.38)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">80</td><td style=\"color: red;font-weight: bold;\">(2.77,7.13)</td><td style=\"color: red;font-weight: bold;\">(3.52,8.69)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">81</td><td style=\"color: red;font-weight: bold;\">(6.02,17.89)</td><td style=\"color: red;font-weight: bold;\">(4.97,12.67)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">86</td><td style=\"color: red;font-weight: bold;\">(1.23,5.07)</td><td style=\"color: red;font-weight: bold;\">(0.98,9.54)</td></tr><tr style=\"color: black;border: 1px solid black;\"><td style=\"color: black;border: 1px solid black;\">89</td><td style=\"color: red;font-weight: bold;\">(-5.77,-0.44)</td><td style=\"color: red;font-weight: bold;\">(-5.46,-0.13)</td></tr></table></center>"
],
"text/plain": [
"<itable.itable.PrettyTable at 0x1170f5cd0>"
]
},
"execution_count": 24,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"CI_table"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "skip"
}
},
"source": [
"## Multiple splits "
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"collapsed": true,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"np.random.seed(100)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"collapsed": true,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"M_est2 = split_glm_group_lasso(loss, epsilon, int(0.5*n), penalty)\n",
"mv2 = multiple_queries([M_est1, M_est2])\n",
"mv2.solve()\n",
"active_union_2splits = M_est1.selection_variable['variables'] + M_est2.selection_variable['variables']"
]
},
{
"cell_type": "code",
"execution_count": 12,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"37\n",
"['P6D' 'P35I' 'P41L' 'P62V' 'P65R' 'P67G' 'P67N' 'P68G' 'P69D' 'P69i'\n",
" 'P75I' 'P75M' 'P75T' 'P77L' 'P83K' 'P90I' 'P102Q' 'P103R' 'P115F' 'P116Y'\n",
" 'P118I' 'P135V' 'P162C' 'P174H' 'P178M' 'P181C' 'P184V' 'P190A' 'P196E'\n",
" 'P202V' 'P207E' 'P210W' 'P211K' 'P215F' 'P215Y' 'P219R' 'P228H']\n"
]
}
],
"source": [
"nactive = np.sum(active_union_2splits)\n",
"active_set_2splits_labels = NRTI_muts[active_union_2splits]\n",
"print(nactive)\n",
"print(active_set_2splits_labels)"
]
},
{
"cell_type": "code",
"execution_count": 39,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"average length 5.53858882054\n"
]
}
],
"source": [
"np.random.seed(100)\n",
"target_sampler_boot, target_observed = glm_target(loss,\n",
" active_union_2splits,\n",
" mv2,\n",
" bootstrap=True)\n",
"\n",
"target_sample_boot = target_sampler_boot.sample(ndraw=20000,\n",
" burnin=0)\n",
"\n",
"LU_ms_boot = target_sampler_boot.confidence_intervals(target_observed,\n",
" sample=target_sample_boot)\n",
"print \"average length\", average_length(LU_ms_boot)"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "skip"
}
},
"source": [
"# Comparing the lengths of CIs\n",
"\n",
"| | <font size=\"5\"> Average length </font> | \n",
"| ------------- |:-------------:|\n",
"| <font size=\"5\">Data splitting </font> | <font size=\"5\"> 6.87 </font> |\n",
"| <font size=\"5\">Data carving (1 split) </font> | <font size=\"5\"> 4.87 </font> |\n",
"| <font size=\"5\">Data carving (2 splits)</font> | <font size=\"5\"> 5.54 </font>|"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## Multiple queries on the data"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"### Query 1: Randomized marginal screening\n",
"- 20 mutations: ['P35I' 'P39A' 'P41L' 'P67N' 'P74I' 'P74V' 'P75I' 'P83K' 'P101E' 'P118I' 'P122E' 'P135V' 'P184V' 'P200A' 'P200I' 'P208Y' 'P210W' 'P211K' 'P215Y' 'P228H']"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## Inference for the selected coefficients \n",
"### after randomized MS"
]
},
{
"cell_type": "code",
"execution_count": 105,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"import pandas as pd\n",
"results1 = pd.read_pickle(\"MS.pkl\")\n",
"active_var1= list(results1.ix[0])\n",
"nactive1 = len(active_var1)\n",
"CI1 = np.zeros((nactive1,2))\n",
"CI1[:,0]= np.array([list(results1.ix[2])[i][0] for i in range(nactive1)])\n",
"CI1[:,1] = np.array([list(results1.ix[2])[i][1] for i in range(nactive1)])\n",
"labels1 = list(results1.ix[4])"
]
},
{
"cell_type": "code",
"execution_count": 117,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "fragment"
}
},
"outputs": [
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAABJQAAAK2CAYAAAAYHsZCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nkiRJkmGRDqUkSZIkSZIkSZIkSZJkWPw/81WLEe58SzMAAAAASUVORK5CYII=\n",
"text/plain": [
"<matplotlib.figure.Figure at 0x119f97e10>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"plot_CI(np.array(labels1), np.array(list(results1.ix[1])), CI1, list(results1.ix[3]), \"Marginal screening\")"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"### Query 2:\n",
"- Add interactions based on 20 mutations selected in MS.\n",
"- Run randomized Lasso on the new dataset.\n",
"- Lasso selects 17 mutations:\n",
" ['P75I' 'P83K' 'P122E' 'P135V' 'P184V' 'P41L:P67N' 'P41L:P118I' 'P67N:P135V' 'P67N:P211K' 'P67N:P215Y' 'P67N:P228H' 'P74V:P208Y' 'P83K:P184V' 'P135V:P200A' 'P184V:P228H' 'P211K:P215Y' 'P211K:P228H'].\n",
" "
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "fragment"
}
},
"source": [
"### Inference for the union of both supports\n",
" - 32 mutations:\n",
" ['P35I' 'P39A' 'P41L' 'P67N' 'P74I' 'P74V' 'P75I' 'P83K' 'P101E' 'P118I'\n",
" 'P122E' 'P135V' 'P184V' 'P200A' 'P200I' 'P208Y' 'P210W' 'P211K' 'P215Y'\n",
" 'P228H' 'P41L:P67N' 'P41L:P118I' 'P67N:P135V' 'P67N:P211K' 'P67N:P215Y'\n",
" 'P67N:P228H' 'P74V:P208Y' 'P83K:P184V' 'P135V:P200A' 'P184V:P228H'\n",
" 'P211K:P215Y' 'P211K:P228H']."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## Inference for the union of selected coefficients\n",
"### after two queries (MS + Lasso)\n",
"\n",
"<!--\n",
"![image](two_steps_randomized_MS_plus_LASSO.pdf)\n",
"-->"
]
},
{
"cell_type": "code",
"execution_count": 118,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "skip"
}
},
"outputs": [],
"source": [
"import pandas as pd\n",
"results2 = pd.read_pickle(\"MS_plus_LASSO.pkl\")\n",
"active_var2= list(results2.ix[0])\n",
"p=126\n",
"nactive2 = len(active_var2)\n",
"CI2 = np.zeros((nactive2,2))\n",
"CI2[:,0]= np.array([list(results2.ix[2])[i][0] for i in range(nactive2)])\n",
"CI2[:,1] = np.array([list(results2.ix[2])[i][1] for i in range(nactive2)])\n",
"labels2 = list(results2.ix[4])"
]
},
{
"cell_type": "code",
"execution_count": 119,
"metadata": {
"collapsed": false,
"slideshow": {
"slide_type": "fragment"
}
},
"outputs": [
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAABJQAAALnCAYAAADfzB0QAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n",
"text/plain": [
"<matplotlib.figure.Figure at 0x11b8a5650>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"plot_CI(np.array(labels2), np.array(list(results2.ix[1])), CI2, list(results2.ix[3]), \"MS + LASSO\")"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"## Selective inference software\n",
"\n",
"- https://github.com/jonathan-taylor/selective-inference\n",
" \n",
"- LASSO, group LASSO, forward-stepwise, marginal screening and more soon!\n",
"\n",
"- GLMs\n",
" \n",
"- including their combination into multiple views / queries on the data\n",
"\n",
"- two samplers: plugin CLT and bootstrap\n",
"\n",
"- either parametric or non-parametric covariance estimates\n",
"\n",
"- cross-validation"
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"- [Tukey (1980)]():\n",
"\n",
"> The picture of the scientist struct - as by lightening - with a question is very far from the truth."
]
},
{
"cell_type": "markdown",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"### References\n",
"\n",
"- Markovic, J., & Taylor, J. (2016). *Bootstrap inference after using multiple queries for model selection.* arXiv preprint arXiv:1612.07811.\n",
"\n",
"- Markovic, J., Xia, L., & Taylor, J. (2017). *Adaptive p-values after cross-validation.* arXiv preprint arXiv:1703.06559.\n",
"\n",
"- Bi, N., Markovic J., Xia, L., & Taylor, J. (2017) *Inferactive data analysis.* arXiv preprint arXiv:1707.06692.\n",
"\n",
"- Harris, X. T., Panigrahi, S., Markovic, J., Bi, N., & Taylor, J. (2016). *Selective sampling after solving a convex problem. arXiv preprint.* arXiv preprint arXiv:1609.05609.\n",
"\n",
"- Tian, X., & Taylor, J. E. (2015). *Selective inference with a randomized response.* The Annals of Statistics, to appear.\t\n",
"\n",
"- Lee, J. D., Sun, D. L., Sun, Y., & Taylor, J. E. (2016). *Exact post-selection inference, with application to the lasso.* The Annals of Statistics, 44(3), 907-927.\n",
"\n",
"- Diaconis, P. (1981). *Magical thinking in the analysis of scientific data.* Annals of the New York Academy of Sciences, 364(1), 236-244.\n",
"\n",
"- Tukey, J. W. (1980). *We need both exploratory and confirmatory.* The American Statistician, 34(1), 23-25.\n",
"\n",
"- Fithian, W., Sun, D., & Taylor, J. (2014). *Optimal inference after model selection.* arXiv preprint arXiv:1410.2597.\n",
"\n",
"- Barber, R. F., & Candès, E. J. (2015). *Controlling the false discovery rate via knockoffs.* The Annals of Statistics, 43(5), 2055-2085."
]
}
],
"metadata": {
"celltoolbar": "Slideshow",
"kernelspec": {
"display_name": "Python 2",
"language": "python",
"name": "python2"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 2
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython2",
"version": "2.7.11"
}
},
"nbformat": 4,
"nbformat_minor": 0
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment