blog_01.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# код для заметки блога \"Scikit-Learn: тонкие вопросы о реализации методов машинного обучения\"\n",
    "\n",
    "https://dyakonov.org/2021/03/03/ml-scikit-learn/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Populating the interactive namespace from numpy and matplotlib\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "%pylab inline\n",
    "plt.style.use('seaborn-dark')\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\") # отключение варнингов\n",
    "pd.set_option('display.max_columns', None) # pd.options.display.max_columns = None \n",
    "# pd.set_option('display.max_rows', None) # не прятать столбцы при выводе дата-фреймов\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib as mpl\n",
    "plt.rc('font', size=14)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [],
   "source": [
    "# для картинки\n",
    "from dj_booklib_v3 import *"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## к вопросу о svm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "предсказания: [0 1]\n",
      "вероятности: [0.64550527 0.35658176]\n",
      "[-0.891277055847369, 0, 0.8912770558473689]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWEAAADDCAYAAABTXKCGAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nO3dfVzN9//H8cfpylXJtUaZbBMm5iKyuRiT+S6FGXOxfK2JKNclJDKmZehLw6RNc01YGWFDaWZmLla53JayIZk1RLWS8/ujn7M11EGnz+mc1/1263bzuTrnddx4enuf9+fzUqnVajVCCCEUYaJ0AUIIYcwkhIUQQkESwkIIoSAJYSGEUJCEsBBCKEhCWAghFGRWHm+iUqkAePvtyYwYEYSZmXl5vK0QRq9nz2pKlyBKUS4j4Us/nmX4oCFs3ryYieO68+uv58vjbYUQQu+pyuNmjT9/ughAzO5deE70AeDVV98iMDBK128thFGTkbD+K9c54X7/cWXTys8ASEjYiouLJbm5d8qzBCGE0Cvl/sVcr1d7kHnqZ822u3t94uKiyrsMIYTQC4qsjjA3NyfrfDpfRK0HICzMl4ULxyCPsRBCGBtFl6h16/QKezZtA2Dv3rX06mXFrVt/KFmSEEKUK8XXCXdo047rZy9otr1HOZOS8q2CFQkhRPlRPIQBTExMyDqfzoFtX2JhXhk/v/8QERFIYWGh0qUJIYRO6UUI3/dSS0cO7/qSt/r0JTp6Cb17W/P775eVLksIIXSmXNcJP45aDo01v3Z19WTixKVlWJEQxkHWCes/vRoJ/1PW+XSO7jkAwK5dn/3/muLbClclhBBlS29DGOB5+yZkpJzHtWcvANzdbeSWZyGEQdHrEAaoZFGJtcsiCA2aA8B777Xjk0+my5piIYRB0Ns54Yc5n/oLnd7oCUCPzv0Y77ecatWql8lrC2GIZE5Y/+n9SPifHJ57nt/PpDJjwhQSDn/JmNGdOHJkt9JlCSHEE6tQI+F/OnL8GG8MfQuA996bw6BBkzAxqVD/pgihczIS1n8VNoQB0n69SDuXbprtLVtSqVmzvk7eS4iKSEJY/1XooaN9o2f541waTZ5tDMCgQc+xalWQskUJAZibg42NihdeQPNjY6PCXJrKiH+p0CEMRa2Tjn2VQMIXOwHYsiWMwYNfID8/T+HKhLGysjLB3h5WrfoRR8ctNG68HkfHLURGJmFvX3RciPsq9HTEv+Xk5jLcdzQHDiUCsGZNCs88Y18u7y0EFI2A7e3h1Vd3kJaW/cBxe3srEhLcuXAB7t7VfT0yHaH/DOqf5KpVqrD10zVM9vahWtXqjPF+mYSErUqXJYxI7doqFi9OfmgAA6SlZRMWlkydOgb1V088BYP8kzBzkj/ffrmLZ+2a8sEHI1i0yEfaKIlyYWWl5rPPzpV4zmefncPK6l45VST0nUGGMEAjWzvit21i0uix7N27Bt8xr5CamqJ0WcIIZGcXlHj81q2SjwvjYrAhDEVtlIImT2X76nXcyclm3LhXiY1dKbc8C52ysip5CUT16rJEQvzNoEP4vm6dXuG7uDjaOnbm44+nMGfmQGmjJHQiO1uFp2ezEs/x9GxGdrZR/NUTWjCaPwl1atVm1/pVzJs+k+9P7Md7lDPJyYeULksYmD/+UDN5civs7a0eetze3opJk1px/brMCYsiRhPCUNRGaeyIkXy1eTsW5pXx93+DtWvnSxslUWYKCiAjw4SEBHcmTHDUTE1Ur27OxImOJCS4k5FhUi7L00TFYFDrhB9H9u3bTH1/Fptjt9OqhTPTZn5O3boNlS5LGAgzM6hTx6TYKojsbBOuX79XrgEs64T1n9GG8H2bY7YzafZMLMwqMcV/BS+/3EfpkoQoMxLC+s+opiMe5u1+b/JN7C7q17Nl9uzBfPzxFLnlWQhRbow+hAGea2xPYuxWvP/rSWzsSsaN7SptlIQQ5cLopyP+7auEA4z29+PmrSzs7V9k5cojqFQqpcsS4onIdIT+k5Hwv/R6tQeHdxZ160hLO80bb9QiIyNdyZKEEAZMQvghnqlfX9NG6e7dAoYPb0lMzCdKlyWEMEAyHVGKDduj8Z3uD0Djxi1YufKItFESFYZMR+g/SZNSDH1zIBeOJgGQnn6GwKlu/PlnpsJVCSEMhYSwFmpYW/PHuTQWz/mA5LPfM9qrI4cO7VC6LCGEAZDpiMd05qfzdHZ7HYBXXulDYOAazM0tFK5KiIeT6Qj9JyPhx9SiqQOp3/8IwLff7uSNN2px6dLPClclhKioJISfQM0aNcg6n86wAYMAGOXVkfj4aIWrEkJURBLCTyF8/gKSDhyi6XOtmD//XRYtGittlIQQj0VC+CnZNbQlfvsmpnj7snfvWny8X5Y2SkIIrUkIlwEzMzMCJ/nxxer15OTeZty4V4mJWSFtlIQQpZIQLkNdO73Md3FxtGvVhWXL/AkOHCBtlIQQJZIQLmP32yh9MD2IoyfjGe0lbZSEEI8m64R1KOn0KTx8xnMp4wIAe/bcwNTUTOGqhDGRdcL6T0bCOtT6xZZ8u/PvO+u8PNtx7dolBSsSQugbCWEds7K0JOt8Ov+b9yG/XUll2LBmLFo0VumyhBB6QkK4nAwfOJiNn3wKwJ49awgPnyxtlIQQEsLl6fXur3Hx+ClcunVnx44IRr7blhMn4pUuSwihoFJD+NSpU0RFRbF7924KCwsfOD5r1iydFGaorCwt2Ryxms0Rq8m49isBAW54eLwoa4qFMFIlro6IiYkhMDCQRo0akZWVRb169Vi1ahU2Njaac1q3bk1SUlKJb2KsqyNKc/TkcXoPHqDZjom5TLVq1gpWJAyNrI7QfyWOhCMjIwkLC2P37t188803tGnTBg8PD/744+8bEGQE9+Q6tGnH72dSNdv9+jVk/fpQBSsSQpS3EkP4ypUr9OzZEwALCwvef/99nJ2dGT16NHl5RV8qSSfip2NqakrW+XR2b9wGQFTUXFxcLLl3757ClQkhykOJIWxjY8OJEyeK7ZszZw61a9dmzJgx5OTk6LQ4Y9KxbTtNGyWAGVP7kJUlbZSEMHQlhvDw4cPx9vZm/fr1f19gYkJ4eDiWlpa4uro+9Ms68WQ0bZTen8+ps0fxHtWRH37Yp3RZQggdMg0ODg5+1MGWLVvSuHFjTE1NcXBw+PsiU1PeeOMN6tSpw40bN+jXr1+Jb5L3x80yK9jQqVQqXmrpiKtLL3btS2DbF8v4668cWrXqjKmpqdLliQqmSRNpvaXv5NkReiwnN5eZIXOJ2rwBh+dfYkbQGho0aKJ0WaICkdUR+k9u1tBjVatUYfH784lauoLLGWmM8X6ZAwe2KF2WEKIMSQhXAO6v/4dvv9yFfaNmhIR4snDhGHJzbytdlhCiDEgIVxCaNkpjxvHVV+twd7fhiy+WK12WEOIpaRXCO3fufOSxmJiYMitGlMzMzIzAiVOIiSparbJ8+VRpoyREBadVCC9evJj+/ftz6NDfHSLS09MZPnw4Cxcu1Flx4uG6OL/MyX3f0L51N5Yt86dXLysuXjyrdFlCiCeg1eqIgoICoqOjWbVqFQ4ODjRv3pzVq1czYMAAJk6ciJWVVYnXy+oI3VCr1dRp3kQzEl64cDetW3dRuCqhT2R1hP57rCVq+fn5fPzxx6xbt44NGzbQrFkzra6TENatrV/G4v/+HG7eygJg167rWFhUVrgqoQ8khPWfViE8YcIEza/VajX79u3DxcVFs2/JkiUlXi8hrHu379yhUdsXNdvr15+jXj1bBSsS+kBCWP9pNSdctWpVzU+1atXo27dvsX1CeZbVqpF1Pp3XunQDwHu0M998E6twVUKI0jz1HXOJiYl07dq1xHNkJFy+LlxMp9egQWTduEblylXZuvUilSpVUbosoQAZCes/rUP49u3bpKamkp+fr9mXmZlJUFAQJ0+eLPFaCeHyl5+fj41jUwCaNGrOjFlrePbZ5gpXJcqbhLD+M9PmpISEBCZNmkRubi4qlUrzbXylSpXo27evTgsUT8bCwoKs8+l8fTCe0X5++Iztgo/vInr3Hi7PgBZCj2g1J7xo0SL8/f05cuQI5ubmHDt2jNWrV9OtWzdGjhyp6xrFU3Dp1p3Du3bzokN7Fi/2IWTOMO7ckafaCaEvtArhS5cuMXToUGrUqIFKpcLS0pJOnToxevRopk+frusaxVOyqVePvZs/J2iyPwe/28WYUZ04e/YHpcsSQqBlCNesWZOMjAwArKysuHixaI63adOmnDlzRnfViTJjamrKpNE+xG3Ywj31PSZNcmHTpsXSRkkIhWkVwm5ubgwYMIDbt2/TtWtXfH19iYiIYMqUKdjZ2em6RlGGOrRpx5G4XXTu8B8+/XSWtFESQmFar46IjY3F3d2dnJwcgoODSUlJwc7ODj8/v2JdNx5GVkfoH7VazZroTUybO4cqlasxdVokTk4upV8oKhRZHaH/pLOGkTv3y88MGzOOtF/PUaWKJZs3p1KlivzFNRQSwvpPqxBesGBBicenTp1a4nEJYf2Wm5fHwJHDOfzDUQA+/zxZ2igZCAlh/afVOuGUlBTNr48fP067du0027LmtOKrUrkyO9dtYdW6z5m7eDFjvF/mtZ5DGD8+TOnShDB4jz0d0apVK5KTkx/rTWQkXHFcunKZVt1f0Wzv2HGVKlUsFaxIPA0ZCeu/x25vJCNfw2bboCGXk85ptt/9b2t+/PGgghUJYdi0CuHc3Fxyc3PJyckBIC8vT7MvNzdXpwWK8lelcmWyzqezY81G/vgzE39/V/r3t5U2SkLogFbTEc2aNdOMgNVq9QOj4bNnS26tI9MRFdfpc2fp0vc/AHRq74Lf9EiqV6+tcFVCWzIdof+0CuGjR4+WeLxDhw4lHpcQrtjUajURa6OY/sEcABYujKN165IfXyr0g4Sw/nuidcJqtZqrV69Sr149TE1NSz1fQtgwLI+KZGbIPM32nj03MDXVaoGNUIiEsP7Tak44MTERJycn3N3dSU9Pp0+fPvTo0YMuXbpw/PhxXdco9MTYESP59cRpzbb/pF5cu/abghUJUfFp/ShLLy8vOnXqhJeXFy+++CIxMTF4eHgwb9680l9AGIz7bZRWfvQ/fkk7jffoThw6tEPpsoSosLSajmjTpg0nT54kOzsbJycnvvnmG+rWrcu9e/do166ddNYwUhcupjN0zHh+Sk3GzW0ko0eHSBslPSPTEfpPq5Fw7dq1OXbsGFZWVsybN4+6desCcOLECWrVqqXTAoX+avJsYxJjtuLj6cWXX0YyfmxXLl4seaWMEKI4rULY19eXjz76CIC33noLgCtXruDl5cXYsWN1V53QexYWFswNCGRzxGqyblzDZ2wX4uJWy5piIbSk9eqIu3fvYmZW/Jvwn3/+mRdeeKHUa2U6wjhcvXYND58pHE/+hm6d3Jg4dQWWljWULsuoyXSE/tMqhH/4oeRWOE5OTiUelxA2Hvfu3WNp5CfMC1vEvXuFeHnNY9CgiUqXZbQkhPWf1nfMVapUiTp16jzw30yVSsX+/ftLvF5C2PgcPXmc3oMHAPDee3MYNGgSJiaP/agS8ZQkhPWfViG8c+dOwsPDcXBwwM/Pj0aNGj3Wm0gIG6ffLl/Cw9eP5DNHANi06Rdq17ZRuCrjIiGs/7SeEy4sLGTz5s1ERkby2muv4ePjQ40a2s33SQgbL7VaTctuncjIvApA9+4DmTFjtcJVGQ8JYf2n9f8PTU1NGTp0KHFxcdSpU4c333yTiIgI8vPzdVmfqOBUKhWnE49weNfXAMTHR7N06SQKCuTPjRCg5Uj4448/fmDfjRs32L59O9bW1sTHx5d4vYyEBRS1UZoV+gGfblgLQFRUEg0bPqdwVYZNRsL6T6unr1y+fPmh+19//fUyLUYYtiqVK/PR7LlkZF4lbv/XjPF+mQkTl/Laa28rXZoQipFuy0IRl65cZtjYSaScPYqLyzDGjVskbZR0QEbC+k/rEI6Li2P//v1cu3YNlUqFjY0NLi4uuLi4lHqthLB4mLt377Jg2VIWLl8KQEBAJD17Dla4KsMiIaz/tH6KWlBQEFD0MJ/WrVtTWFjItGnTCAuTjrziyZiZmTFjwmQ2fvIpAKGhI9m+fZnc8iyMilYj4fbt2/PZZ5/RqlWrYvuTkpLw9PQs9ZnCMhIWpfkjK4vh4wL47tjXOLfrid/0SKyt6yhdVoUnI2H9p9VI2NzcnObNmz+wv0WLFlhYWJR5UcL41K5Vi53rIggJnM3xpERGj3ImKSlR6bKE0DmtQvjdd99l2bJlFBYWavYVFhayatUqRowYoavahJFRqVSMHv4uX0dvp0rlqvj7uxIVNZfCwrtKlyaEzmg1HTF8+HBOny5qa2Nra8u9e/fIyMhArVbTpEmTYnN4W7dufeB6mY4Qj+v2nTsEzJ3Fxi+20bKZE9OD1lCvnp3SZVU4Mh2h/574Zo1H8fX1fWCfhLB4Ulu/jGViUCAmJqZMnrKcLl36Kl1ShSIhrP9knbDQe2m/XmSo93jOpybRoMFzLFmyjxo16ipdVoUgIaz/tArhCRMmlHh8yZIlJR6XEBZPKz8/n9kfhbByTdHDf4KDN/LKK24KV6X/JIT1n1ZfzH399ddUrVr1kT9C6JqFhQUhgbOZ7O0DQHDwEEaN6ihrikWFp9VIuFWrViQnJz/xm8hIWJSly1czcOzWSbMdHZ0m0xOPICNh/afVSFilUum6DiG01tDmGa6fvaDZHjjQnsOHdypYkRBPTqunqBUWFnLw4MFHHu/WrVuZFSSENkxMTMg6n87WL2MZ5TeB2bOLnjmxd+8taaMkKhSte8w98gVUKs6ePVvi9TIdIXTp39MT0kbpbzIdof9kiZowCGq1ml5v9+d40o8AzJu3jY4d5XnXEsL6T0JYGJRtO2PxmvL3ksq4uCzMzY33+SYSwvpPJs+EQRnQpy9Xks9ptieOe5XLl1MVrEiIkkkIC4NTuVJlss6ns3bZSjIyf2WM98vs27dJ6bKEeCgJYWGwXHu+zrc743je/kVCQ0eyYMFocnNvK12WEMVICAuDZvtMA/Zv3YDf2PHs27eBsaNf5ueff1S6LCE0JISFwbvfRmnHmo3k/ZXLhPE9pI2S0BsSwsJovNLBme927aL9S91YsSKAWdP7c/PmdaXLEkZOQlgYlfttlD6cGcyR4/t4663GrF07X+myhBGTdcLCaEVt3sDkWTM023v23MDUVKs7+SsMWSes/2QkLIzWiLeHcvbQUc127941SE8/o2BFwhhJCAujVr9uPbLOp+PUpi0AXl4dCAhwV7gqURHcuHEDHx8fHBwcuHTp0hO/joSwEMDeTdtZtywCgBMnDuDiYklu7h2Fq9Ifubl3uXYtl4yMHK5dyyU317g7YP/yyy/0798fKyurp34tCWEh/t8bPXtxNeUnzba7e32ZngAyM3OYMeMozz+/iQYN1vH885sIDPyBzMwcnb2ng4MD8fHxABw8eJCOHTty+3bRjTbLli2jefPmODo64ujoSLNmzQgNDQWKHuQUFhZG9+7dadOmDX369NG8DhQ9lnfx4sV07tyZ9u3b4+Pjw/Xr10s99m/Xr1/nww8/xNPT86k/q4SwEP9gYWFB1vl03nQt6l/n5dWBoUMdjHZNcWZmDp06xfK//50iO7sAgOzsAsLCUujUKVanQXzf8uXL8fb2xtLSEoArV67Qv39/UlJSSElJwcnJSXNubGwsmzdvZu3atRw/fpwhQ4YwefJkbt26BcDatWvZs2cPGzduJDExEYDAwMBSj/2bs7MzHTt2LJPPJyEsxENELg7XfGn3+++X6dXLitu3byhcVfnKzb1LaGgSaWnZDz2elpbNggVJ5OXpbmri4MGDXLt2jWHDhmn2Xbp0CVtb24ee7+bmxtdff42trS0mJia4urqSk5NDamrRQ5y2b9/O22+/jZ2dHVWrVmXGjBkMHDiw1GO6JCEsxCPUr1uvWBsl71HOnDnzvYIVla/s7AIiI8+VeE5k5Dlu3SrQyfvfvHmTDz/8kICAACws/n4c6W+//UajRo0eek1ubi4hISF07twZR0dHunTpAhR1675/bcOGDTXnN2zYkJ49e5Z6TJckhIUowf02Sl9t+QITlQmTJvViw4aPKCwsVLo0nSssVGumIB7l1q0C7t3TzVRNcHAwly5domnTppp9eXl5ZGRk8Nxzzz30mjlz5pCUlMSaNWtITk7m8OHDxY6rVCru3bv30GtLOqZLEsJCaKF96zZ8F7eTbp1cWb16DtP9Xbl+PUPpsnTK1FSFlZV5iedUr26OiYluGgGHhYUxZMgQZs+erdn3008/YW5u/sgQTk5Oxt3dnSZNmqBSqTh16lSx43Z2dqSlpWm2L1++TFRUFGq1usRjuiQhLISWqltVZ+unS1gyL5SzPx3He3RHjh79SumydMbKypyRIx/dXxJg5MhmVK9eclA/jfHjx5OWlsb27dsB2LlzJx06dCg2PfFPdnZ2nDp1ivz8fE6fPs2GDRuwsLAgMzMTgAEDBrBp0yZ++eUXcnNzWbx4MYcOHUKlUpV4TJcM6x5NIXRMpVLhMfBtOrRtx7Ax4wkMfJMBA3zx9JyDhUUlpcsrU1WqmBEQ0JqYmPSHfjlnb2/F1KmtqVxZdzFiaWnJtGnTmDt3LseOHePQoUMsX778kef7+fnh7++Pk5MTLVq0ICQkhBo1ahAUFIS1tTUeHh7cunWLd955h8LCQjp27Mj8+UXPDinp2L/NnDmT2NhYzSi5d+/eqFQq5s6dS79+/R7rM8qzI4R4Qnl/5TErdD6R69fwQhNHZgStxdb2eaXLKqYsnh2RmZlDaGgSkZHnyM4uoHr1ohHy1KmtqV+/ahlUadwMLoSXLllE/s3SlxJZWNdg/IQp5VCRMHS79u1lbEAAd+8WMH7CEnr2HKx0SRpl9QCfvLy7mi/hTExUVK9urtMRsDExuN/F/Js3+KBr51LPC0w8VA7VCGNQ1EbJkWFjJhEaOpLQ0JFs3Pgzdeo8o3RpZaZyZTMJXR2RL+aEKANFbZTWa7aHDHmB7duXKViRqCgkhIUoI2ZmZmSdT2fWlKkArFgRgIfHi0Z7y7PQjoSwEGVs4qix/HLkJABXr15kZkBfbtz4XeGqhL6SEBZCB2rVrMkf59IIDZrD0ZMHGDjQnpMnE5QuS+ghCWEhdESlUuH1zn/5cGYwAAEBbqxePYfCQuN+Fq8oTkJYCB0b5TGC306eYeibA9mw4SP6utvIqFhoSAgLUQ6qVa1K+PwFrFq8lL/y85g6tQ+ffDJN6bKEHpAQFqIcDXB1J259NADbtn1MSIgnf/2Vq3BVQkkGt/rawrqGVjdiWFjXKIdqhHiQc3snfj1xGjePtzlwYAsXfk5iRtAa7O1fVLo0oaX8/HxCQkKIj48nNzeXNm3aMGfOHOrXr//Yr2Vwty0LUZEcOJSI15Qp/Hnjd+rXb8TatafL9KldT3Pb8oczZpD3iB5r/1S5Th2mPeJBN4YqNDSU7777jvDwcGrUqMH8+fNJTU1ly5Ytj/1aMh0hhIJ6dO7Kd7viAMjM/JV5s4dw61aWwlUVybt+neD27Uv90SaoH5c+N/osLCwkOjqasWPHYmdnh5WVFf7+/iQnJ3P27NnH/qwSwkIorF6dulw/e4Fgv2kkfreTAQMacfr0EaXL0hv61ujz4sWLZGdn06JFC82+WrVqYWNjQ0pKymN/PglhIfSAiYkJ4728CRg3EYDJk19n/foFRtFGqST62Ojzxo2ipzRaW1sX229tbc2ff/752J9RQlgIPRLgO5H04yl069SHqKj3eWdoM86fP650WYqoaI0+1Wr1E83nSwgLoWeqW1qx9dP/sfSDBVzPysDXtxvR0UuULqvc6Wujz1q1agE8MOq9efMmNWvWLPX6f5MQFkIPqVQq3nlrEBtWRAIQERFIWJgv+fl/KVxZ+dHXRp92dnZYW1sXe+3MzEyuXr3KSy+99NifU0JYCD3Wu0dPUr//kYbPNCAuLgpX19ocPLhd6bLKjT42+jQ1NWXw4MGsWLGCS5cucevWLRYsWICzszMvvPDCY39Gg7tZQwhDU7NGDVISDhO37yve8RnFvHnDyc/Pw8VlqNKl6Zy+NvocN24cOTk5vPPOO+Tl5dGhQwfCwsKe6DPKzRpCVCCXr2YweNQ4Tp8/BkBsbAZVq1o98vynuVkjeNQogtu3L/28Y8cIjoh44vcxdjISFqICaWjzDAlfbKZui6I50b59nyEgIFInzUUr16lD8LFjWp0nnpyMhIWooHyn+7Nhe9HDgLy9P6R//7GYmBT/mqesui0L3ZEQFqICy/rzT/47bhrf/rCXJo2aMzN4PXZ2fy/pkhDWf7I6QogKrFbNmuxY+wkLZr3PhV/P4unZlkOHdihdlngMMhIWwkC8vyiU/0Ws0Gzv3v0nvXvLI1v1nYSwEAbkTk4Odm3+frBMOfz1Fk9JpiOEMCDVqlYl63w6qxYvVboUoaVyGQkLIYR4OBkJCyGEgiSEhRBCQRLCQgihIAlhIYRQkISw0AuFhYUsXbqUFi1aaB5bKIQxkAf4CMXl5eUxYsQI6tWrR5UqVZQuR4hyJSPhCsLBwYGWLVtq2nw3b96c8ePHAxAeHo6HhwcrV67E2dmZtm3bsnDhQs1C/WnTphEeHq55rZ07d+Lg4MD3338PQExMDK+99hovvfQSbm5uHDx4EIDvv/8eBwcH7ty5o7nWw8ND6/bi06ZN09SYn59P9+7d2bHjwVtqc3JycHV1ZenSpZiZybhAGBcJ4QokPDxc0+a7b9++xY6dPn2anJwcEhMTWb16NRs2bCAuLu6B17h79y5Lly7VdCYoKChg/fr1hISEcPLkSdzc3JgyZYpWvbZKay/+T1u3bqV69eq4ubk9cKxWrVp4eHho+9sghEGREDYQarUaX19fLCwsaN26Nd27d2f//v0PnLdlyxYcHByoW7cuAObm5kRHR9OhQwdu377N9evXsbGxeeCRiA9TWnvx+/Lz84mIiGDy5MlP1I1WCEMm//czELa2tpibm2u2GzRowMmTJ4udk5uby6pVq1i9ejWenp7FjkVHRzNz5kxsbW1ZtWpVsWPOzs6aXxcUFAxeSJ4AAAHmSURBVNCyZUvN64WEhJCYmMjNmzc159xvLw5F/ziEh4fj4OBAt27dnv6DCmFgZCRsIP49faBWqx8YdX7++ef06NGDxo0bP3D9wIEDSUxMpG3btvj6+lJYWKg5duTIEc00iJOTk2Z/ae3FAfbt28e6deto1qzZU35CIQyThLCByMjIoKCgQLN95coV6tevr9m+efMmmzZtwsfH54FrL1++DED9+vV57733SE1N5erVq6W+Z2ntxQFcXFxYv349UVFRXLhw4Uk+mhAGTULYQKjValauXEl+fj5JSUkkJCTQs2dPzfFt27YxePBgatWqVey6n376CVdXV3744QcKCwuJj4+nZs2a1KtXr9T3LK29+H0tWrRgwIABzJ49u2w+rBAGRELYQDRp0gRTU1O6du3KiBEjGDZsGL1799Yct7KyYsSIEQ9c17RpUwIDAwkICKBdu3bs2bOH8PDwYvPLj+Ln50d6ejpOTk7MmzePKVOm0K9fP4KCgjTL3O6bOHEiqampD70RIyYmRrP07saNGwQFBeHo6MjMmTMf/zdCiApGHmVpAMLDw4mPj5c7zYSogGQkLIQQCpIQFkIIBcl0hBBCKEhGwkIIoSAJYSGEUJCEsBBCKEhCWAghFCQhLIQQCpIQFkIIBf0fE3/CcSfgOJAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 288x216 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "X = [[0, 0], [1, 1]] # обучение \n",
    "y = [0, 1]\n",
    "X2 = X.copy() # контроль\n",
    "\n",
    "from sklearn.svm import SVC\n",
    "svm = SVC(kernel='rbf', gamma=1.0, probability=True)\n",
    "svm.fit(X, y)\n",
    "\n",
    "a = svm.predict(X2)\n",
    "p = svm.predict_proba(X2)\n",
    "print (f'предсказания: {a}')\n",
    "print (f'вероятности: {p[:,1]}')\n",
    "\n",
    "\n",
    "\n",
    "X = np.array(X)\n",
    "y = np.array(y)\n",
    "\n",
    "plt.figure(figsize=(4, 3))\n",
    "plot_2d_separator(svm, X, fill=True, twocolor=True);\n",
    "show_scatter(X[:, 0], X[: ,1], y, figsize=(4, 3.5),\n",
    "             size=100, eps=0.2, newfig=False, show_legend=True);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [
    {
     "ename": "ValueError",
     "evalue": "Cannot have number of splits n_splits=5 greater than the number of samples: n_samples=2.",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-90-94caf026725a>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0mcalibrated_clf\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mCalibratedClassifierCV\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbase_estimator\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msvm\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 4\u001b[0;31m \u001b[0mcalibrated_clf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      5\u001b[0m \u001b[0mcalibrated_clf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpredict_proba\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/lib/python3.7/site-packages/sklearn/calibration.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, X, y, sample_weight)\u001b[0m\n\u001b[1;32m    180\u001b[0m                     \u001b[0msample_weight\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_check_sample_weight\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msample_weight\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mX\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    181\u001b[0m                 \u001b[0mbase_estimator_sample_weight\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msample_weight\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 182\u001b[0;31m             \u001b[0;32mfor\u001b[0m \u001b[0mtrain\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtest\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mcv\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msplit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    183\u001b[0m                 \u001b[0mthis_estimator\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mclone\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbase_estimator\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    184\u001b[0m                 \u001b[0;32mif\u001b[0m \u001b[0mbase_estimator_sample_weight\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/lib/python3.7/site-packages/sklearn/model_selection/_split.py\u001b[0m in \u001b[0;36msplit\u001b[0;34m(self, X, y, groups)\u001b[0m\n\u001b[1;32m    331\u001b[0m                 (\"Cannot have number of splits n_splits={0} greater\"\n\u001b[1;32m    332\u001b[0m                  \" than the number of samples: n_samples={1}.\")\n\u001b[0;32m--> 333\u001b[0;31m                 .format(self.n_splits, n_samples))\n\u001b[0m\u001b[1;32m    334\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    335\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mtrain\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtest\u001b[0m \u001b[0;32min\u001b[0m \u001b[0msuper\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msplit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mgroups\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mValueError\u001b[0m: Cannot have number of splits n_splits=5 greater than the number of samples: n_samples=2."
     ]
    }
   ],
   "source": [
    "from sklearn.calibration import CalibratedClassifierCV\n",
    "\n",
    "calibrated_clf = CalibratedClassifierCV(base_estimator=svm)\n",
    "calibrated_clf.fit(X, y)\n",
    "calibrated_clf.predict_proba(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "предсказания: [0 1]\n",
      "вероятности: [0.08333339 0.91680141]\n"
     ]
    }
   ],
   "source": [
    "X = [[0, 0], [1, 1]] # обучение \n",
    "y = [0, 1]\n",
    "X2 = X.copy() # контроль\n",
    "\n",
    "X = X * 10 # дублируем объекты\n",
    "y = y * 10\n",
    "\n",
    "from sklearn.svm import SVC\n",
    "svm = SVC(kernel='rbf', gamma=1.0, probability=True)\n",
    "svm.fit(X, y)\n",
    "\n",
    "a = svm.predict(X2)\n",
    "p = svm.predict_proba(X2)\n",
    "print (f'предсказания: {a}')\n",
    "print (f'вероятности: {p[:,1]}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "class sklearn.linear_model.Ridge(alpha=1.0, # Regularization strength\n",
    "                                 fit_intercept=True, normalize=False,\n",
    "                                 copy_X=True, max_iter=None, tol=0.001,\n",
    "                                 solver='auto', random_state=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "class sklearn.svm.SVC(C=1.0, # Regularization parameter\n",
    "                      kernel='rbf', degree=3, gamma='scale', coef0=0.0,\n",
    "                      shrinking=True, probability=False, tol=0.001,\n",
    "                      cache_size=200, class_weight=None, verbose=False,\n",
    "                      max_iter=- 1, decision_function_shape='ovr',\n",
    "                      break_ties=False, random_state=None)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## к вопросу о lgb"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((100000, 20), (100000,))"
      ]
     },
     "execution_count": 92,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X = np.random.rand(100000, 20)\n",
    "y = (X[:,0] + X[:,1])*(X[:,2] - X[:, 3])\n",
    "X.shape, y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 3 µs, sys: 0 ns, total: 3 µs\n",
      "Wall time: 6.2 µs\n",
      "rmse = 0.02065529588662666\n"
     ]
    }
   ],
   "source": [
    "%time\n",
    "import lightgbm as lgb\n",
    "\n",
    "params = {'objective': 'regression',\n",
    "          'boosting_type': 'gbdt',\n",
    "          'learning_rate': 0.1}\n",
    "\n",
    "cv_results = lgb.cv(params,\n",
    "                    lgb.Dataset(X, y),\n",
    "                    num_boost_round=100,\n",
    "                    nfold=10, # cv.split(X), # 10,\n",
    "                    metrics='rmse',\n",
    "                    stratified=False\n",
    "                    )\n",
    "\n",
    "print (f\"rmse = {cv_results['rmse-mean'][99]}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0, 0.4)"
      ]
     },
     "execution_count": 94,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x216 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(7, 3))\n",
    "plt.plot(cv_results['rmse-mean'], lw=2, c='#000099')\n",
    "plt.xlabel('итерации')\n",
    "plt.ylabel('RMSE')\n",
    "plt.grid(lw=2)\n",
    "plt.xlim([0, 100])\n",
    "plt.ylim([0, 0.4])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 2 µs, sys: 0 ns, total: 2 µs\n",
      "Wall time: 5.96 µs\n",
      "rmse = 0.020629265953391353\n"
     ]
    }
   ],
   "source": [
    "%time\n",
    "\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.model_selection import KFold\n",
    "\n",
    "model = lgb.LGBMRegressor() # модель\n",
    "cv = KFold(n_splits=10, random_state=1) # способ контроля\n",
    "\n",
    "scores = cross_val_score(model, X, y, cv=cv, scoring='neg_root_mean_squared_error')\n",
    "print (f'rmse = {-mean(scores)}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LGBMRegressor(boosting_type='gbdt', class_weight=None, colsample_bytree=1.0,\n",
       "              importance_type='split', learning_rate=0.1, max_depth=-1,\n",
       "              min_child_samples=20, min_child_weight=0.001, min_split_gain=0.0,\n",
       "              n_estimators=100, n_jobs=-1, num_leaves=31, objective=None,\n",
       "              random_state=None, reg_alpha=0.0, reg_lambda=0.0, silent=True,\n",
       "              subsample=1.0, subsample_for_bin=200000, subsample_freq=0)"
      ]
     },
     "execution_count": 96,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lgb.LGBMRegressor()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 7 µs, sys: 1 µs, total: 8 µs\n",
      "Wall time: 6.2 µs\n",
      "rmse = 0.02065176102281662\n"
     ]
    }
   ],
   "source": [
    "%time\n",
    "import lightgbm as lgb\n",
    "# параметры модели\n",
    "params = {'objective': 'regression',\n",
    "          'boosting_type': 'gbdt',\n",
    "          'learning_rate': 0.1}\n",
    "# контроль\n",
    "cv_results = lgb.cv(params,\n",
    "                    lgb.Dataset(X, y),\n",
    "                    num_boost_round=100,\n",
    "                    folds=cv,\n",
    "                    metrics='rmse',\n",
    "                    # stratified=False\n",
    "                    )\n",
    "\n",
    "print (f\"rmse = {cv_results['rmse-mean'][-1]}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Пример с дисбалансом"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 train: [1 2 5 6 7 8 9] test: [0 3 4]\n",
      "1 train: [0 2 3 4 7 8 9] test: [1 5 6]\n",
      "2 train: [0 1 3 4 5 6 8 9] test: [2 7]\n",
      "3 train: [0 1 2 3 4 5 6 7] test: [8 9]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import StratifiedKFold\n",
    "\n",
    "skf = StratifiedKFold(n_splits=4, random_state=1, shuffle=False)\n",
    "x = np.arange(10)\n",
    "y = [1,1,1,0,0,0,0,0,0,0]\n",
    "\n",
    "for i, (train_index, test_index) in enumerate(skf.split(x, y)):\n",
    "    print (i, 'train:', train_index, 'test:', test_index)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}