diff --git a/player-permutation.ipynb b/player-permutation.ipynb index d02f8b8..04797c1 100644 --- a/player-permutation.ipynb +++ b/player-permutation.ipynb @@ -2,7 +2,7 @@ "cells": [ { "cell_type": "code", - "execution_count": 19, + "execution_count": 7, "metadata": {}, "outputs": [ { @@ -21,6 +21,7 @@ "from csgo_wp.model import LR_CNN\n", "from sklearn.metrics import log_loss, roc_auc_score, accuracy_score\n", "import torch\n", + "from itertools import permutations\n", "\n", "import matplotlib.pyplot as plt\n", "%matplotlib inline\n", @@ -28,42 +29,6 @@ "import warnings\n", "warnings.filterwarnings('ignore')\n", "\n", - "def test(model, loader, device):\n", - " model.eval()\n", - " model.to(device)\n", - "\n", - " targets = []\n", - " outputs = []\n", - "\n", - " with torch.no_grad():\n", - " for index, (data, target) in enumerate(loader):\n", - " targets.append(target)\n", - "\n", - " # permute the data - just the 2nd and 4th T players\n", - " data[:, 0] = data[:, 0, [0, 3, 2, 1, 4], :]\n", - " data[:, 0] = data[:, 0, :, [0, 3, 2, 1, 4]]\n", - "\n", - " data[:, 2] = data[:, 2, [0, 3, 2, 1, 4], :]\n", - " data[:, 3] = data[:, 3, :, [0, 3, 2, 1, 4]]\n", - "\n", - " data[:, 4] = data[:, 4, :, [0, 3, 2, 1, 4]]\n", - " data[:, 4] = data[:, 4, [0, 3, 2, 1, 4], :]\n", - "\n", - " data = data.to(device)\n", - " output = model(data)\n", - " outputs.append(output)\n", - "\n", - " y_pred = torch.cat(outputs, dim=0).cpu().numpy().astype(float)\n", - " y_true = torch.cat(targets, dim=0).cpu().numpy().astype(float)\n", - "\n", - " print('\\n' + '-' * 30)\n", - " print('Results')\n", - " print(f'Accuracy: {accuracy_score(y_true, y_pred > 0.5):.4f}')\n", - " print(f'AUC: {roc_auc_score(y_true, y_pred):.4f}')\n", - " print(f'Log loss: {log_loss(y_true, y_pred):.4f}')\n", - "\n", - " return\n", - "\n", "test_dataset = CSGODataset(transform=transform_multichannel,\n", " dataset_split='test',\n", " verbose=False,\n", @@ -93,36 +58,144 @@ }, { "cell_type": "code", - "execution_count": 20, + "execution_count": 9, + "metadata": {}, + "outputs": [], + "source": [ + "model.to(device)\n", + "\n", + "outputs = []\n", + "\n", + "with torch.no_grad():\n", + " for index, (data, target) in enumerate(test_loader):\n", + " model_output = []\n", + " for permutation in permutations([0, 1, 2, 3, 4], r=5):\n", + " # permute the data\n", + " data[:, 0] = data[:, 0, permutation, :]\n", + " data[:, 0] = data[:, 0, :, permutation]\n", + "\n", + " data[:, 2] = data[:, 2, permutation, :]\n", + " data[:, 3] = data[:, 3, :, permutation]\n", + "\n", + " data[:, 4] = data[:, 4, permutation, :]\n", + " data[:, 4] = data[:, 4, :, permutation]\n", + "\n", + " # get the model output\n", + " data = data.to(device)\n", + " output = model(data)\n", + " model_output.append(output)\n", + " outputs.append(model_output)" + ] + }, + { + "cell_type": "code", + "execution_count": 29, "metadata": {}, "outputs": [ { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "------------------------------\n", - "Results\n", - "Accuracy: 0.6963\n", - "AUC: 0.8290\n", - "Log loss: 0.5143\n" - ] + "data": { + "text/plain": [ + "0 0.010496\n", + "1 0.009052\n", + "2 0.004919\n", + "3 0.003570\n", + "4 0.003872\n", + "dtype: float64" + ] + }, + "execution_count": 29, + "metadata": {}, + "output_type": "execute_result" } ], "source": [ - "# permuted test data\n", - "test(model=model,\n", - " loader=test_loader,\n", - " device=device,\n", - " )" + "stds = []\n", + "\n", + "for perm in outputs:\n", + " permutation_results = pd.DataFrame([x.detach().cpu().numpy() for x in perm]).T\n", + " stds.append(permutation_results.std(axis=1))\n", + " \n", + "stds = pd.concat(stds, axis=0)\n", + "stds.head()" ] }, { "cell_type": "code", - "execution_count": null, + "execution_count": 54, "metadata": {}, - "outputs": [], - "source": [] + "outputs": [ + { + "data": { + "text/plain": [ + "Text(0, 0.5, 'Incidence')" + ] + }, + "execution_count": 54, + "metadata": {}, + "output_type": "execute_result" + }, + { + "data": { + "image/png": "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\n", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "import seaborn as sns\n", + "\n", + "plt.figure(figsize=(10, 7))\n", + "sns.violinplot(permutation_results.iloc[15])\n", + "plt.xlabel('Assigned win probability')\n", + "plt.ylabel('Incidence')" + ] + }, + { + "cell_type": "code", + "execution_count": 51, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "0.010396631729692322" + ] + }, + "execution_count": 51, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "permutation_results.iloc[15].std()" + ] + }, + { + "cell_type": "code", + "execution_count": 38, + "metadata": {}, + "outputs": [ + { + "data": { + "image/png": "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\n", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "plt.figure(figsize=(10, 7))\n", + "stds.hist(bins=300)\n", + "plt.xlabel('Standard deviation of predictions')\n", + "plt.ylabel('Incidence');" + ] } ], "metadata": {