Commit a4e52106 authored by Bannier Delphine's avatar Bannier Delphine
Browse files

update evaluation

parent 61dd2e62
......@@ -1275,20 +1275,6 @@
"### Test set"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": 21,
......@@ -1316,13 +1302,6 @@
"#print('competition score :', score)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": 22,
......@@ -1379,7 +1358,7 @@
},
{
"cell_type": "code",
"execution_count": 24,
"execution_count": 10,
"metadata": {},
"outputs": [],
"source": [
......@@ -1424,7 +1403,7 @@
},
{
"cell_type": "code",
"execution_count": 25,
"execution_count": 11,
"metadata": {},
"outputs": [
{
......@@ -1433,7 +1412,7 @@
"(140, 240, 240)"
]
},
"execution_count": 25,
"execution_count": 11,
"metadata": {},
"output_type": "execute_result"
}
......@@ -1443,15 +1422,16 @@
]
},
{
"cell_type": "code",
"execution_count": null,
"cell_type": "markdown",
"metadata": {},
"outputs": [],
"source": []
"source": [
"# Train score\n",
"https://medium.com/hal24k-techblog/how-to-generate-neural-network-confidence-intervals-with-keras-e4c0b78ebbdf"
]
},
{
"cell_type": "code",
"execution_count": 31,
"execution_count": 52,
"metadata": {},
"outputs": [
{
......@@ -1465,7 +1445,7 @@
"name": "stdout",
"output_type": "stream",
"text": [
"[<tf.Tensor 'input_3_3:0' shape=(None, 240, 240, 1) dtype=float32>]\n"
"[<tf.Tensor 'input_3_8:0' shape=(None, 240, 240, 1) dtype=float32>]\n"
]
}
],
......@@ -1473,7 +1453,8 @@
"from keras.models import load_model\n",
"import tensorflow as tf\n",
"tf.compat.v1.disable_eager_execution()\n",
"dropout = 0.3\n",
"\n",
"dropout = 0.2\n",
"num_iter = 20\n",
"input_data=trainImagesX\n",
"input_data = np.asarray(input_data).reshape(140,240,240,1)\n",
......@@ -1491,44 +1472,192 @@
},
{
"cell_type": "code",
"execution_count": 32,
"execution_count": 53,
"metadata": {},
"outputs": [],
"source": [
"FVC_pred = sc.inverse_transform(predictions)"
]
},
{
"cell_type": "code",
"execution_count": 54,
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"import matplotlib.pyplot as plt\n",
"plt.hist(np.std(FVC_pred, axis = 1), bins=15)\n",
"plt.show()"
]
},
{
"cell_type": "code",
"execution_count": 55,
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"diff_fvc = (np.mean(FVC_pred,axis=1) - sc.inverse_transform(trainY))\n",
"diff = (np.mean(predictions,axis=1) - trainY)\n",
"plt.hist(diff_fvc, bins=30)\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Score\n"
]
},
{
"cell_type": "code",
"execution_count": 56,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"MSE : 706541.3749511036\n",
"RMSE : 840.560155462477\n"
]
}
],
"source": [
"MSE = np.mean(diff_fvc*diff_fvc)\n",
"RMSE = np.sqrt(MSE)\n",
"print(\"MSE : \",MSE)\n",
"print(\"RMSE : \", RMSE)"
]
},
{
"cell_type": "code",
"execution_count": 57,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"MSE : 0.9764751854626166\n",
"RMSE : 0.9881675897653275\n"
]
}
],
"source": [
"MSE = np.mean(diff*diff)\n",
"RMSE = np.sqrt(MSE)\n",
"print(\"MSE : \",MSE)\n",
"print(\"RMSE : \", RMSE)"
]
},
{
"cell_type": "code",
"execution_count": 58,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[ 0.34668085, 0.06881715, 0.19557495, ..., 0.34579813,\n",
" 0.20636405, 0.37848622],\n",
" [ 0.40555188, 0.40567875, 0.95923394, ..., -0.49564478,\n",
" -0.09881385, -0.39867502],\n",
" [-0.2788513 , 0.04609472, -0.43507397, ..., 0.56579775,\n",
" 0.48428255, 0.02170943],\n",
" ...,\n",
" [-0.29737601, -0.11840692, -0.23768747, ..., -0.11680555,\n",
" -0.39183071, -0.20751245],\n",
" [ 0.05204046, -0.08606576, 0.44020665, ..., 0.65810513,\n",
" -0.14248499, 0.0195321 ],\n",
" [ 0.40544796, 0.38101956, 0.47285724, ..., -0.37723538,\n",
" 0.08306865, -0.27934054]])"
"-13.872223340763814"
]
},
"execution_count": 32,
"execution_count": 58,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"predictions"
"from postprocessing.evaluate import compute_score\n",
"\n",
"compute_score(trainY,np.mean(FVC_pred,axis=1),np.std(FVC_pred,axis=1))"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Test score"
]
},
{
"cell_type": "code",
"execution_count": 39,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"(36, 240, 240)"
]
},
"execution_count": 39,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"testImagesX.shape"
]
},
{
"cell_type": "code",
"execution_count": 59,
"metadata": {},
"outputs": [],
"source": [
"input_data=testImagesX\n",
"input_data = np.asarray(input_data).reshape(36,240,240,1)\n",
"num_samples = input_data.shape[0]\n",
"\n",
"predictions = np.zeros((num_samples, num_iter))\n",
"for i in range(num_iter):\n",
" predictions[:,i] = predict_with_dropout([input_data,1])[0].reshape(-1)"
]
},
{
"cell_type": "code",
"execution_count": 33,
"execution_count": 60,
"metadata": {},
"outputs": [],
"source": [
"FVC_pred = sc.inverse_transform(predictions)"
]
},
{
"cell_type": "code",
"execution_count": 61,
"metadata": {},
"outputs": [
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAAD4CAYAAADiry33AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAARWklEQVR4nO3df6wlZX3H8fenKH9UabTuBWEBl7aEdjVCyXX9QWu0FrO7GqmtadkYpdZkq5FGE03ctok1aZrQNNqmxUq2laCJBW10lYSVH7FNqPEXF7IgFChbsoZ1CXvRFjSa6uq3f9zZcL2es/ecM+fee/bh/UpOzsw8zzPzPQP72dk5Z2ZSVUiS2vVzG12AJGltGfSS1DiDXpIaZ9BLUuMMeklq3DM2uoBBNm3aVFu2bNnoMiTppHHnnXc+XlVzg9pmMui3bNnCwsLCRpchSSeNJN8c1uapG0lqnEEvSY0z6CWpcQa9JDXOoJekxhn0ktS4VYM+yTlJ/j3J/UnuS/LubvkvJrktyUPd+3OHjN+e5MEkB5PsmfYHkCSd2ChH9MeA91bVrwEvA96VZCuwB/hiVZ0PfLGb/ylJTgE+AuwAtgK7urGSpHWyatBX1aNVdVc3/V3gfmAzcBnw8a7bx4HfGTB8G3Cwqh6uqh8CN3TjJEnrZKwrY5NsAX4d+BpwRlU9Ckt/GSQ5fcCQzcAjy+YPAy8dsu7dwG6Ac889d5yyNOO27Llp6us8dNXrprq+tahxmqb9efX0MvKXsUmeDXwGeE9VPTnqsAHLBj7Sqqr2VtV8Vc3PzQ28XYMkaQIjBX2SZ7IU8p+sqs92ix9LcmbXfiZwdMDQw8A5y+bPBo5MXq4kaVyj/OomwMeA+6vqw8uabgSu6KavAD4/YPgdwPlJzktyKnB5N06StE5GOaK/BHgL8FtJDnSvncBVwKVJHgIu7eZJclaS/QBVdQy4EriFpS9xP11V963B55AkDbHql7FV9SUGn2sHeM2A/keAncvm9wP7Jy1QktSPV8ZKUuMMeklqnEEvSY0z6CWpcQa9JDXOoJekxhn0ktQ4g16SGmfQS1LjDHpJapxBL0mNM+glqXEGvSQ1zqCXpMYZ9JLUOINekhq36oNHklwLvB44WlUv6pZ9Crig6/Ic4H+r6qIBYw8B3wV+DByrqvkp1S1JGtGqQQ9cB1wNfOL4gqr6g+PTST4EPHGC8a+uqscnLVCS1M8ojxK8PcmWQW3dg8N/H/it6ZYlSZqWvufofxN4rKoeGtJewK1J7kyyu+e2JEkTGOXUzYnsAq4/QfslVXUkyenAbUkeqKrbB3Xs/iLYDXDuuef2LEuSdNzER/RJngH8LvCpYX2q6kj3fhTYB2w7Qd+9VTVfVfNzc3OTliVJWqHPqZvfBh6oqsODGpM8K8lpx6eB1wL39tieJGkCqwZ9kuuBrwAXJDmc5O1d0+WsOG2T5Kwk+7vZM4AvJbkb+DpwU1XdPL3SJUmjGOVXN7uGLP/DAcuOADu76YeBC3vWJ0nqqe+XsZoBW/bcNNX1HbrqdVNdn6SN5S0QJKlxBr0kNc6gl6TGGfSS1DiDXpIaZ9BLUuMMeklqnEEvSY0z6CWpcQa9JDXOoJekxhn0ktQ4g16SGmfQS1LjDHpJapxBL0mNG+VRgtcmOZrk3mXLPpjkW0kOdK+dQ8ZuT/JgkoNJ9kyzcEnSaEY5or8O2D5g+d9W1UXda//KxiSnAB8BdgBbgV1JtvYpVpI0vlWDvqpuB74zwbq3AQer6uGq+iFwA3DZBOuRJPXQ5xz9lUnu6U7tPHdA+2bgkWXzh7tlAyXZnWQhycLi4mKPsiRJy00a9B8Ffhm4CHgU+NCAPhmwrIatsKr2VtV8Vc3Pzc1NWJYkaaWJgr6qHquqH1fVT4B/Yuk0zUqHgXOWzZ8NHJlke5KkyU0U9EnOXDb7RuDeAd3uAM5Pcl6SU4HLgRsn2Z4kaXLPWK1DkuuBVwGbkhwG/gJ4VZKLWDoVcwj4467vWcA/V9XOqjqW5ErgFuAU4Nqqum9NPoUkaahVg76qdg1Y/LEhfY8AO5fN7wd+5qeXkqT145WxktQ4g16SGmfQS1LjDHpJapxBL0mNM+glqXEGvSQ1zqCXpMYZ9JLUOINekhpn0EtS4wx6SWqcQS9JjTPoJalxBr0kNc6gl6TGGfSS1LhVgz7JtUmOJrl32bK/SfJAknuS7EvynCFjDyX5RpIDSRamWbgkaTSjHNFfB2xfsew24EVV9WLgv4A/PcH4V1fVRVU1P1mJkqQ+Vg36qrod+M6KZbdW1bFu9qvA2WtQmyRpCqZxjv6PgC8MaSvg1iR3Jtl9opUk2Z1kIcnC4uLiFMqSJEHPoE/y58Ax4JNDulxSVRcDO4B3JXnlsHVV1d6qmq+q+bm5uT5lSZKWmTjok1wBvB54c1XVoD5VdaR7PwrsA7ZNuj1J0mQmCvok24H3A2+oqu8P6fOsJKcdnwZeC9w7qK8kae2M8vPK64GvABckOZzk7cDVwGnAbd1PJ6/p+p6VZH839AzgS0nuBr4O3FRVN6/Jp5AkDfWM1TpU1a4Biz82pO8RYGc3/TBwYa/qJEm9rRr00izasuemjS5hXa3F5z101eumvk7NJm+BIEmNM+glqXEGvSQ1zqCXpMYZ9JLUOINekhpn0EtS4wx6SWqcQS9JjfPKWP2Mp9tVp1LrPKKXpMYZ9JLUOINekhpn0EtS4wx6SWqcQS9JjRvlUYLXJjma5N5ly34xyW1JHurenztk7PYkDyY5mGTPNAuXJI1mlCP664DtK5btAb5YVecDX+zmf0qSU4CPADuArcCuJFt7VStJGtuqQV9VtwPfWbH4MuDj3fTHgd8ZMHQbcLCqHq6qHwI3dOMkSeto0nP0Z1TVowDd++kD+mwGHlk2f7hbNlCS3UkWkiwsLi5OWJYkaaW1/DI2A5bVsM5Vtbeq5qtqfm5ubg3LkqSnl0mD/rEkZwJ070cH9DkMnLNs/mzgyITbkyRNaNKgvxG4opu+Avj8gD53AOcnOS/JqcDl3ThJ0joa5eeV1wNfAS5IcjjJ24GrgEuTPARc2s2T5Kwk+wGq6hhwJXALcD/w6aq6b20+hiRpmFVvU1xVu4Y0vWZA3yPAzmXz+4H9E1cnSerNK2MlqXEGvSQ1zqCXpMYZ9JLUOINekhpn0EtS4wx6SWqcQS9JjTPoJalxBr0kNc6gl6TGGfSS1DiDXpIaZ9BLUuMMeklq3Kr3o9d0bdlz00aXIOlpxiN6SWrcxEGf5IIkB5a9nkzynhV9XpXkiWV9PtC/ZEnSOCY+dVNVDwIXASQ5BfgWsG9A1/+oqtdPuh1JUj/TOnXzGuC/q+qbU1qfJGlKphX0lwPXD2l7eZK7k3whyQuHrSDJ7iQLSRYWFxenVJYkqXfQJzkVeAPwrwOa7wJeUFUXAv8AfG7Yeqpqb1XNV9X83Nxc37IkSZ1pHNHvAO6qqsdWNlTVk1X1vW56P/DMJJumsE1J0oimEfS7GHLaJsnzk6Sb3tZt79tT2KYkaUS9LphK8vPApcAfL1v2DoCqugZ4E/DOJMeAHwCXV1X12aYkaTy9gr6qvg88b8Wya5ZNXw1c3WcbkqR+vDJWkhpn0EtS4wx6SWqcQS9JjTPoJalxBr0kNc6gl6TGGfSS1DiDXpIaZ9BLUuN8OPgqfJi3pJOdR/SS1DiDXpIaZ9BLUuMMeklqnEEvSY0z6CWpcb2CPsmhJN9IciDJwoD2JPn7JAeT3JPk4j7bkySNbxq/o391VT0+pG0HcH73einw0e5dkrRO1vrUzWXAJ2rJV4HnJDlzjbcpSVqmb9AXcGuSO5PsHtC+GXhk2fzhbtnPSLI7yUKShcXFxZ5lSZKO6xv0l1TVxSydonlXkleuaM+AMTVoRVW1t6rmq2p+bm6uZ1mSpON6BX1VHenejwL7gG0ruhwGzlk2fzZwpM82JUnjmTjokzwryWnHp4HXAveu6HYj8Nbu1zcvA56oqkcnrlaSNLY+v7o5A9iX5Ph6/qWqbk7yDoCqugbYD+wEDgLfB97Wr1xJ0rgmDvqqehi4cMDya5ZNF/CuSbchSerPK2MlqXEGvSQ1zqCXpMYZ9JLUuOaeGeszXqXRTPvPyqGrXjfV9Wl6PKKXpMYZ9JLUOINekhpn0EtS4wx6SWqcQS9JjTPoJalxBr0kNc6gl6TGGfSS1DiDXpIaZ9BLUuP6PDP2nCT/nuT+JPclefeAPq9K8kSSA93rA/3KlSSNq8/dK48B762qu7qHhN+Z5Laq+s8V/f6jql7fYzuSpB4mPqKvqker6q5u+rvA/cDmaRUmSZqOqZyjT7IF+HXgawOaX57k7iRfSPLCE6xjd5KFJAuLi4vTKEuSxBSCPsmzgc8A76mqJ1c03wW8oKouBP4B+Nyw9VTV3qqar6r5ubm5vmVJkjq9gj7JM1kK+U9W1WdXtlfVk1X1vW56P/DMJJv6bFOSNJ4+v7oJ8DHg/qr68JA+z+/6kWRbt71vT7pNSdL4+vzq5hLgLcA3khzolv0ZcC5AVV0DvAl4Z5JjwA+Ay6uqemxTkjSmiYO+qr4EZJU+VwNXT7oNSSePp+PDxk+Wz+yVsZLUOINekhpn0EtS4wx6SWqcQS9JjTPoJalxBr0kNc6gl6TGGfSS1Lg+t0CQpDUz7atOn848opekxhn0ktQ4g16SGmfQS1LjDHpJapxBL0mNM+glqXF9Hw6+PcmDSQ4m2TOgPUn+vmu/J8nFfbYnSRpfn4eDnwJ8BNgBbAV2Jdm6otsO4PzutRv46KTbkyRNps8R/TbgYFU9XFU/BG4ALlvR5zLgE7Xkq8BzkpzZY5uSpDH1uQXCZuCRZfOHgZeO0Gcz8OjKlSXZzdJRP8D3kjzYozaATcDjPdexEax7fVn3+rLuE8hf9xr+gmENfYI+A5bVBH2WFlbtBfb2qOenN5wsVNX8tNa3Xqx7fVn3+rLujdHn1M1h4Jxl82cDRyboI0laQ32C/g7g/CTnJTkVuBy4cUWfG4G3dr++eRnwRFX9zGkbSdLamfjUTVUdS3IlcAtwCnBtVd2X5B1d+zXAfmAncBD4PvC2/iWPbGqngdaZda8v615f1r0BUjXwlLkkqRFeGStJjTPoJalxJ2XQj3DrhV9N8pUk/5fkfSvaDiX5RpIDSRbWr+qR6n5zd6uIe5J8OcmFo46d4bpneX9f1tV8IMlCkt8YdewM1z2z+3tZv5ck+XGSN407di30rHvD9vdYquqkerH0xe9/A78EnArcDWxd0ed04CXAXwHvW9F2CNg0o3W/AnhuN70D+NqoY2ex7pNgfz+bp76nejHwwEmyvwfWPev7e1m/f2PphxpvOhn297C6N3J/j/s6GY/oV731QlUdrao7gB9tRIFDjFL3l6vqf7rZr7J03cFIY2e07o00St3fq+5PK/AsnrqYb9b397C6N9Ko++xPgM8ARycYuxb61H3SOBmDfthtFUZVwK1J7uxuu7Bexq377cAXJhw7TX3qhhnf30nemOQB4Cbgj8YZu0b61A0zvL+TbAbeCFwz7tg11Kdu2Lj9PZY+t0DYKCPfVmGIS6rqSJLTgduSPFBVt0+pthMZue4kr2YpMI+fe+37mfvoUzfM+P6uqn3AviSvBP4S+O1Rx66RPnXDbO/vvwPeX1U/Tn6q+6zv72F1w8bt77GcjEHf67YKVXWkez+aZB9L/3Rbj/8wI9Wd5MXAPwM7qurb44xdI33qnvn9fVxV3Z7kl5NsGnfslE1cd1U9PuP7ex64oQvLTcDOJMdGHLtWJq67qj63gft7PBv9JcG4L5b+cnoYOI+nvjx54ZC+H2TZl7Esnc88bdn0l4Hts1I3cC5LVxG/YtLPPGN1z/r+/hWe+lLzYuBbLB3hzfr+Hlb3TO/vFf2v46kvY2d6f5+g7g3b3+O+Troj+hrh1gtJng8sAL8A/CTJe1h6OMomlv65C0v/gf+lqm6elbqBDwDPA/6xq/FYVc0PGzvrdQNnMNv7+/dYuhfTj4AfAH9QS39qZ31/D6w7yazv77HGznrdbOD/3+PyFgiS1LiT8Vc3kqQxGPSS1DiDXpIaZ9BLUuMMeklqnEEvSY0z6CWpcf8POfTT2hLiIkoAAAAASUVORK5CYII=\n",
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
......@@ -1541,18 +1670,18 @@
],
"source": [
"import matplotlib.pyplot as plt\n",
"plt.hist(np.std(predictions, axis = 1), bins=15)\n",
"plt.hist(np.std(FVC_pred, axis = 1), bins=15)\n",
"plt.show()"
]
},
{
"cell_type": "code",
"execution_count": 43,
"execution_count": 62,
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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\n",
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
......@@ -1564,8 +1693,9 @@
}
],
"source": [
"diff = (np.mean(predictions,axis=1) - trainY)\n",
"plt.hist(diff, bins=30)\n",
"diff_fvc = (np.mean(FVC_pred,axis=1) - sc.inverse_transform(testY))\n",
"diff = (np.mean(predictions,axis=1) - testY)\n",
"plt.hist(diff_fvc, bins=30)\n",
"plt.show()"
]
},
......@@ -1573,20 +1703,41 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"# Score\n"
"## Score\n"
]
},
{
"cell_type": "code",
"execution_count": 48,
"execution_count": 63,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"MSE : 0.9820405205578085\n",
"RMSE : 0.9909795762566495\n"
"MSE : 573115.7800189949\n",
"RMSE : 757.0441070499096\n"
]
}
],
"source": [
"MSE = np.mean(diff_fvc*diff_fvc)\n",
"RMSE = np.sqrt(MSE)\n",
"print(\"MSE : \",MSE)\n",
"print(\"RMSE : \", RMSE)"
]
},
{
"cell_type": "code",
"execution_count": 64,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"MSE : 0.792074402754304\n",
"RMSE : 0.8899856194087093\n"
]
}
],
......@@ -1599,10 +1750,23 @@
},
{
"cell_type": "code",
"execution_count": null,
"execution_count": 65,
"metadata": {},
"outputs": [],
"source": []
"outputs": [
{
"data": {
"text/plain": [
"-13.935864066355839"
]
},
"execution_count": 65,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"compute_score(testY,np.mean(FVC_pred,axis=1),np.std(FVC_pred,axis=1))"
]
}
],
"metadata": {
......
......@@ -41,11 +41,10 @@ def evaluate_mlp(model,df, trainAttrX, trainY,sc):
print("mean difference : {:.2f}%, std: {:.2f}%".format(mean, std))
return preds
def compute_score(y_true, y_pred):
sigma = ( y_true - y_pred ) #########
def compute_score(y_true, y_pred, sigma):
fvc_pred = y_pred
sigma_clip = np.maximum(sigma, 70)
delta = np.minimum(abs(y_true - fvc_pred),1000)
sq2 = math.sqrt(2)
metric = -(delta / sigma_clip)*sq2 - np.log(sigma_clip* sq2)
return (sigma, np.mean(metric))
\ No newline at end of file
return (np.mean(metric))
\ No newline at end of file
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment