KNN.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "authorship_tag": "ABX9TyNqCp3SombUqd6uP4aoIyNO",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/Running-Machine-ML23-2/problem_3/blob/main/KNN.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "e9pVRcEJ32QB"
      },
      "outputs": [],
      "source": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xD3K3AaWE-1Q"
      },
      "source": [
        "# 데이터 불러오기"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "G4sWIuUm76sS"
      },
      "source": [
        "Thanks for 서현, 데이터를 불러올 수 있도록 여러 개로 균등/랜덤하게 나눈 것을 불러옵니다."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "I09DpoEfBPCd",
        "outputId": "4aec6627-7774-4739-cf86-91674fd7f92e"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "id": "gO3-Hn0zExkx"
      },
      "outputs": [],
      "source": [
        "import pickle\n",
        "import numpy as np\n",
        "from collections import Counter\n",
        "\n",
        "# Load the subsetted pickle file  9500개임\n",
        "subset_pickle_path = '/content/drive/MyDrive/dataset_/mon_data_sub1.pkl'\n",
        "with open(subset_pickle_path, 'rb') as f:\n",
        "    X1, X2, y = pickle.load(f)\n",
        "\n",
        "# Convert tuples to lists\n",
        "X1_mon = list(X1) # Array for timestamps\n",
        "X2_mon = list(X2) # Array for packetsize\n",
        "y = list(y) # label\n",
        "\n",
        "outgoing_packets_fraction = []  # Array to store the outgoing packets fraction for unmonitored data\n",
        "std_outgoing_packet_ordering = []  # Array to store the standard deviation of outgoing packet ordering for unmonitored data\n",
        "\n",
        "for i in range(len(X1_mon)):\n",
        "    outgoing_packets = 0\n",
        "    packet_ordering = []\n",
        "    for j in range(len(X1_mon[i])):\n",
        "        if X2_mon[i][j] > 0:\n",
        "            outgoing_packets += 1\n",
        "            packet_ordering.append(X1_mon[i][j])\n",
        "    outgoing_packets_fraction.append(outgoing_packets / len(X1_mon[i]))\n",
        "    std_outgoing_packet_ordering.append(np.std(packet_ordering))\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1GFylnIp8qzI",
        "outputId": "3a67f867-4edd-4c79-f2de-f4c1530a91ab"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Loading datafile...\n"
          ]
        }
      ],
      "source": [
        "import pickle\n",
        "import numpy as np\n",
        "\n",
        "TOTAL_URLS = 3000  # total number in the dataset\n",
        "\n",
        "print(\"Loading datafile...\")\n",
        "with open('./drive/MyDrive/dataset_/unmon_data_sub1.pkl', 'rb') as f:  # Path to unmon_standard10.pkl in Colab\n",
        "    x = pickle.load(f)\n",
        "\n",
        "X1_unmon = [] # Array to store instances (timestamps) - 10,000 instances, e.g., [[0.0, 0.5, 3.4, ...], [0.0, 4.5, ...], [0.0, 1.5, ...], ... [... ,45.8]]\n",
        "X2_unmon = [] # Array to store instances (direction*size) - size information\n",
        "outgoing_packets_fraction_un = []  # Array to store the outgoing packets fraction for unmonitored data\n",
        "std_outgoing_packet_ordering_un = []  # Array to store the standard deviation of outgoing packet ordering for unmonitored data\n",
        "\n",
        "for i in range(TOTAL_URLS):\n",
        "    size_seq = []\n",
        "    time_seq = []\n",
        "    outgoing_packets = 0\n",
        "    packet_ordering = []\n",
        "    for c in x[i]:\n",
        "        dr = 1 if c > 0 else -1\n",
        "        time_seq.append(abs(c))\n",
        "        size_seq.append(dr * 512)  # In the pickle file, there is no size information, so the conversion code is set to multiply by 512 uniformly.\n",
        "        if dr == 1:\n",
        "            outgoing_packets += 1\n",
        "            packet_ordering.append(abs(c))\n",
        "    X1_unmon.append(time_seq)\n",
        "    X2_unmon.append(size_seq)\n",
        "    outgoing_packets_fraction_un.append(outgoing_packets / len(x[i]))\n",
        "    std_outgoing_packet_ordering_un.append(np.std(packet_ordering))\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "id": "QVilwVNBud-2"
      },
      "outputs": [],
      "source": [
        "#Categorical features-3\n",
        "def cat3_feat(X2):\n",
        "  in_packet_of_total_packet=[]\n",
        "  for i in range(len(X2)):\n",
        "    num_in_pack=0\n",
        "    for j in range(len(X2[i])):\n",
        "      if X2[i][j] == -512:\n",
        "        num_in_pack+=1\n",
        "    in_packet_of_total_packet.append(num_in_pack/len(X2[i]))\n",
        "  return in_packet_of_total_packet\n",
        "\n",
        "# 5.number of outgoing packets\n",
        "def cat5_feat(X2_mon):\n",
        "  out_sum=[0]*len(X2_mon)\n",
        "  for i in range(len(X2_mon)):\n",
        "    sum=0\n",
        "    for j in X2_mon[i]:\n",
        "      if j > 0:\n",
        "        sum+=1\n",
        "    out_sum[i]=sum\n",
        "  return out_sum\n",
        "\n",
        "cat5_result=cat5_feat(X2_mon)\n",
        "\n",
        "MAX_SEQUENCE_LENGTH = 100  # 시퀀스의 최대 길이\n",
        "\n",
        "def pad_sequence(seq, max_length):\n",
        "    if len(seq) < max_length:\n",
        "        seq += [0] * (max_length - len(seq))  # 패딩\n",
        "    else:\n",
        "        seq = seq[:max_length]  # 잘라냄\n",
        "    return seq\n",
        "\n",
        "def compute_bursts(sequence):\n",
        "    bursts = []\n",
        "    current_burst = sequence[0]\n",
        "\n",
        "    for size in sequence[1:]:\n",
        "        if np.sign(size) == np.sign(current_burst):\n",
        "            current_burst += size\n",
        "        else:\n",
        "            bursts.append(current_burst)\n",
        "            current_burst = size\n",
        "    bursts.append(current_burst)\n",
        "    return bursts\n",
        "\n",
        "X2_all = X2_mon + X2_unmon\n",
        "\n",
        "X1_all= X1_mon + X1_unmon\n",
        "\n",
        "X2_all = [pad_sequence(x, MAX_SEQUENCE_LENGTH) for x in X2_all]\n",
        "X1_all = [pad_sequence(x, MAX_SEQUENCE_LENGTH) for x in X1_all]\n",
        "\n",
        "X_mon_bursts = [pad_sequence(compute_bursts(x), MAX_SEQUENCE_LENGTH) for x in X2_mon]\n",
        "X_unmon_bursts = [pad_sequence(compute_bursts(x), MAX_SEQUENCE_LENGTH) for x in X2_unmon]\n",
        "X_bursts_all = X_mon_bursts+X_unmon_bursts\n",
        "\n",
        "X_mon_cumsum = [pad_sequence(np.cumsum(x).tolist(), MAX_SEQUENCE_LENGTH) for x in X2_mon]\n",
        "X_unmon_cumsum = [pad_sequence(np.cumsum(x).tolist(), MAX_SEQUENCE_LENGTH) for x in X2_unmon]\n",
        "X_cumsum_all = X_mon_cumsum + X_unmon_cumsum\n",
        "\n",
        "num_packets = [sum(1 for size in instance_sizes if size < 0) for instance_sizes in X2_mon]\n",
        "num_packets_un = [sum(1 for size in instance_sizes if size < 0) for instance_sizes in X2_unmon]\n",
        "X_CAT1 = num_packets+num_packets_un\n",
        "X_CAT2 = outgoing_packets_fraction+outgoing_packets_fraction_un\n",
        "X_CAT3 = cat3_feat(X2_mon)+cat3_feat(X2_unmon)\n",
        "X_CAT4 = std_outgoing_packet_ordering + std_outgoing_packet_ordering_un\n",
        "X_CAT5 = cat5_feat(X2_mon)+cat5_feat(X2_unmon)\n",
        "\n",
        "X_CAT1 = np.array(X_CAT1).reshape(-1, 1)\n",
        "X_CAT2 = np.array(X_CAT2).reshape(-1, 1)\n",
        "X_CAT3 = np.array(X_CAT3).reshape(-1, 1)\n",
        "X_CAT4 = np.array(X_CAT4).reshape(-1, 1)\n",
        "X_CAT5 = np.array(X_CAT5).reshape(-1, 1)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "id": "qIAusSwHsnNL"
      },
      "outputs": [],
      "source": [
        "y = y + [-1] * 3000"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.neighbors import KNeighborsClassifier\n",
        "from sklearn.metrics import accuracy_score\n",
        "from itertools import combinations\n",
        "from sklearn.model_selection import train_test_split\n",
        "import numpy as np\n",
        "\n",
        "# 피처 목록\n",
        "features = [X2_all, X1_all, X_bursts_all, X_cumsum_all, X_CAT1, X_CAT2, X_CAT4,X_CAT3,X_CAT5]\n",
        "feature_names = ['X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4','X_CAT3','X_CAT5']\n",
        "\n",
        "# 결과를 저장할 딕셔너리\n",
        "results = {}\n",
        "\n",
        "# 가능한 모든 조합에 대해\n",
        "for r in range(1, len(features) + 1):\n",
        "    for subset_indices in combinations(range(len(features)), r):\n",
        "        subset = [features[i] for i in subset_indices]\n",
        "        subset_names = [feature_names[i] for i in subset_indices]\n",
        "\n",
        "        # 피처를 합칩니다\n",
        "        X = np.concatenate([np.array(feature) for feature in subset], axis=1)\n",
        "        X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)\n",
        "\n",
        "        # 모델 초기화\n",
        "        clf = KNeighborsClassifier(n_neighbors=3)\n",
        "\n",
        "        # 모델 학습\n",
        "        clf.fit(X_train, y_train)\n",
        "\n",
        "        # 예측\n",
        "        y_pred = clf.predict(X_test)\n",
        "\n",
        "        # 성능 평가\n",
        "        accuracy = accuracy_score(y_test, y_pred)\n",
        "\n",
        "        # 결과 저장\n",
        "        results[tuple(subset_names)] = accuracy\n",
        "\n",
        "# 결과 출력\n",
        "for subset, accuracy in results.items():\n",
        "    print(\"Features: \", subset)\n",
        "    print(\"Accuracy: {:.3f}\".format(accuracy))\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rNGDKYiN4iRl",
        "outputId": "b7f6a02f-65af-43b8-d4b5-063e28abc950"
      },
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Features:  ('X2_all',)\n",
            "Accuracy: 0.726\n",
            "Features:  ('X1_all',)\n",
            "Accuracy: 0.310\n",
            "Features:  ('X_bursts_all',)\n",
            "Accuracy: 0.327\n",
            "Features:  ('X_cumsum_all',)\n",
            "Accuracy: 0.644\n",
            "Features:  ('X_CAT1',)\n",
            "Accuracy: 0.182\n",
            "Features:  ('X_CAT2',)\n",
            "Accuracy: 0.158\n",
            "Features:  ('X_CAT4',)\n",
            "Accuracy: 0.139\n",
            "Features:  ('X_CAT3',)\n",
            "Accuracy: 0.160\n",
            "Features:  ('X_CAT5',)\n",
            "Accuracy: 0.161\n",
            "Features:  ('X2_all', 'X1_all')\n",
            "Accuracy: 0.730\n",
            "Features:  ('X2_all', 'X_bursts_all')\n",
            "Accuracy: 0.334\n",
            "Features:  ('X2_all', 'X_cumsum_all')\n",
            "Accuracy: 0.657\n",
            "Features:  ('X2_all', 'X_CAT1')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X_CAT2')\n",
            "Accuracy: 0.750\n",
            "Features:  ('X2_all', 'X_CAT4')\n",
            "Accuracy: 0.740\n",
            "Features:  ('X2_all', 'X_CAT3')\n",
            "Accuracy: 0.750\n",
            "Features:  ('X2_all', 'X_CAT5')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X1_all', 'X_bursts_all')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X1_all', 'X_cumsum_all')\n",
            "Accuracy: 0.648\n",
            "Features:  ('X1_all', 'X_CAT1')\n",
            "Accuracy: 0.276\n",
            "Features:  ('X1_all', 'X_CAT2')\n",
            "Accuracy: 0.310\n",
            "Features:  ('X1_all', 'X_CAT4')\n",
            "Accuracy: 0.321\n",
            "Features:  ('X1_all', 'X_CAT3')\n",
            "Accuracy: 0.310\n",
            "Features:  ('X1_all', 'X_CAT5')\n",
            "Accuracy: 0.280\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X_bursts_all', 'X_CAT1')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X_bursts_all', 'X_CAT2')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X_bursts_all', 'X_CAT4')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X_bursts_all', 'X_CAT3')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X_bursts_all', 'X_CAT5')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X_cumsum_all', 'X_CAT1')\n",
            "Accuracy: 0.677\n",
            "Features:  ('X_cumsum_all', 'X_CAT2')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X_cumsum_all', 'X_CAT4')\n",
            "Accuracy: 0.652\n",
            "Features:  ('X_cumsum_all', 'X_CAT3')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X_cumsum_all', 'X_CAT5')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X_CAT1', 'X_CAT2')\n",
            "Accuracy: 0.211\n",
            "Features:  ('X_CAT1', 'X_CAT4')\n",
            "Accuracy: 0.209\n",
            "Features:  ('X_CAT1', 'X_CAT3')\n",
            "Accuracy: 0.211\n",
            "Features:  ('X_CAT1', 'X_CAT5')\n",
            "Accuracy: 0.372\n",
            "Features:  ('X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.162\n",
            "Features:  ('X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.165\n",
            "Features:  ('X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.232\n",
            "Features:  ('X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.164\n",
            "Features:  ('X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.188\n",
            "Features:  ('X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.233\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all')\n",
            "Accuracy: 0.659\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT1')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT2')\n",
            "Accuracy: 0.730\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT4')\n",
            "Accuracy: 0.734\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT3')\n",
            "Accuracy: 0.730\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT5')\n",
            "Accuracy: 0.758\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT1')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT2')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT4')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT3')\n",
            "Accuracy: 0.334\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT5')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT1')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT2')\n",
            "Accuracy: 0.672\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT4')\n",
            "Accuracy: 0.664\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT3')\n",
            "Accuracy: 0.672\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT5')\n",
            "Accuracy: 0.673\n",
            "Features:  ('X2_all', 'X_CAT1', 'X_CAT2')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X_CAT1', 'X_CAT4')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X_CAT1', 'X_CAT3')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X_CAT1', 'X_CAT5')\n",
            "Accuracy: 0.764\n",
            "Features:  ('X2_all', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.740\n",
            "Features:  ('X2_all', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.750\n",
            "Features:  ('X2_all', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.740\n",
            "Features:  ('X2_all', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT1')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT2')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT4')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT3')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT5')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT1')\n",
            "Accuracy: 0.677\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT2')\n",
            "Accuracy: 0.648\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT4')\n",
            "Accuracy: 0.651\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT3')\n",
            "Accuracy: 0.648\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT5')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X1_all', 'X_CAT1', 'X_CAT2')\n",
            "Accuracy: 0.276\n",
            "Features:  ('X1_all', 'X_CAT1', 'X_CAT4')\n",
            "Accuracy: 0.283\n",
            "Features:  ('X1_all', 'X_CAT1', 'X_CAT3')\n",
            "Accuracy: 0.276\n",
            "Features:  ('X1_all', 'X_CAT1', 'X_CAT5')\n",
            "Accuracy: 0.412\n",
            "Features:  ('X1_all', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.320\n",
            "Features:  ('X1_all', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.310\n",
            "Features:  ('X1_all', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.280\n",
            "Features:  ('X1_all', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.320\n",
            "Features:  ('X1_all', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.289\n",
            "Features:  ('X1_all', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.280\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT1')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT2')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT4')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT3')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT5')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X_bursts_all', 'X_CAT1', 'X_CAT2')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X_bursts_all', 'X_CAT1', 'X_CAT4')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X_bursts_all', 'X_CAT1', 'X_CAT3')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X_bursts_all', 'X_CAT1', 'X_CAT5')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X_bursts_all', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X_bursts_all', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X_bursts_all', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X_bursts_all', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X_bursts_all', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X_bursts_all', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X_cumsum_all', 'X_CAT1', 'X_CAT2')\n",
            "Accuracy: 0.677\n",
            "Features:  ('X_cumsum_all', 'X_CAT1', 'X_CAT4')\n",
            "Accuracy: 0.677\n",
            "Features:  ('X_cumsum_all', 'X_CAT1', 'X_CAT3')\n",
            "Accuracy: 0.677\n",
            "Features:  ('X_cumsum_all', 'X_CAT1', 'X_CAT5')\n",
            "Accuracy: 0.678\n",
            "Features:  ('X_cumsum_all', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.652\n",
            "Features:  ('X_cumsum_all', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.661\n",
            "Features:  ('X_cumsum_all', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X_cumsum_all', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.652\n",
            "Features:  ('X_cumsum_all', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X_cumsum_all', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X_CAT1', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.211\n",
            "Features:  ('X_CAT1', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.211\n",
            "Features:  ('X_CAT1', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.372\n",
            "Features:  ('X_CAT1', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.210\n",
            "Features:  ('X_CAT1', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.379\n",
            "Features:  ('X_CAT1', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.371\n",
            "Features:  ('X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.166\n",
            "Features:  ('X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.189\n",
            "Features:  ('X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.233\n",
            "Features:  ('X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.190\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT1')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT2')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT4')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT3')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT5')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT1')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT2')\n",
            "Accuracy: 0.659\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT4')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT3')\n",
            "Accuracy: 0.659\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT5')\n",
            "Accuracy: 0.672\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT1', 'X_CAT2')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT1', 'X_CAT4')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT1', 'X_CAT3')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT1', 'X_CAT5')\n",
            "Accuracy: 0.764\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.734\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.730\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.758\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.734\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.758\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT4')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT3')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT5')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT1', 'X_CAT4')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT1', 'X_CAT3')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT1', 'X_CAT5')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.334\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT3')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT5')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.664\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.671\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.673\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.664\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.673\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.673\n",
            "Features:  ('X2_all', 'X_CAT1', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X_CAT1', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X_CAT1', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.764\n",
            "Features:  ('X2_all', 'X_CAT1', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X_CAT1', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.764\n",
            "Features:  ('X2_all', 'X_CAT1', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.764\n",
            "Features:  ('X2_all', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.741\n",
            "Features:  ('X2_all', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT4')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT3')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT5')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT4')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT3')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT5')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2')\n",
            "Accuracy: 0.677\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4')\n",
            "Accuracy: 0.677\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT3')\n",
            "Accuracy: 0.677\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT5')\n",
            "Accuracy: 0.678\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.651\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.648\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.651\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X1_all', 'X_CAT1', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.283\n",
            "Features:  ('X1_all', 'X_CAT1', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.276\n",
            "Features:  ('X1_all', 'X_CAT1', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.412\n",
            "Features:  ('X1_all', 'X_CAT1', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.283\n",
            "Features:  ('X1_all', 'X_CAT1', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.413\n",
            "Features:  ('X1_all', 'X_CAT1', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.412\n",
            "Features:  ('X1_all', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.320\n",
            "Features:  ('X1_all', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.289\n",
            "Features:  ('X1_all', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.280\n",
            "Features:  ('X1_all', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.289\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT3')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT5')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X_bursts_all', 'X_CAT1', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X_bursts_all', 'X_CAT1', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X_bursts_all', 'X_CAT1', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X_bursts_all', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X_bursts_all', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X_bursts_all', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X_bursts_all', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.677\n",
            "Features:  ('X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.677\n",
            "Features:  ('X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.678\n",
            "Features:  ('X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.677\n",
            "Features:  ('X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.678\n",
            "Features:  ('X_cumsum_all', 'X_CAT1', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.678\n",
            "Features:  ('X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.652\n",
            "Features:  ('X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X_cumsum_all', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X_cumsum_all', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.211\n",
            "Features:  ('X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.379\n",
            "Features:  ('X_CAT1', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.371\n",
            "Features:  ('X_CAT1', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.379\n",
            "Features:  ('X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.192\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT4')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT3')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT5')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT4')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT3')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT5')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT3')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT5')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.659\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.672\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.672\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.672\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT1', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT1', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT1', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.764\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT1', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT1', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.764\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT1', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.764\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.734\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.758\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT3')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT5')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT1', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT1', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT1', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.664\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.673\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.673\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.673\n",
            "Features:  ('X2_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.764\n",
            "Features:  ('X2_all', 'X_CAT1', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.764\n",
            "Features:  ('X2_all', 'X_CAT1', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.764\n",
            "Features:  ('X2_all', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT3')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT5')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.677\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.677\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.678\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.677\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.678\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.678\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.651\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X1_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.283\n",
            "Features:  ('X1_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.413\n",
            "Features:  ('X1_all', 'X_CAT1', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.412\n",
            "Features:  ('X1_all', 'X_CAT1', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.413\n",
            "Features:  ('X1_all', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.289\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X_bursts_all', 'X_CAT1', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X_bursts_all', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.677\n",
            "Features:  ('X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.678\n",
            "Features:  ('X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.678\n",
            "Features:  ('X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.678\n",
            "Features:  ('X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.379\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT3')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT5')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.672\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.672\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.672\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.764\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT1', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.764\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT1', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.764\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.759\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT1', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.673\n",
            "Features:  ('X2_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.764\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.327\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.677\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.678\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.678\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.678\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.662\n",
            "Features:  ('X1_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.413\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.678\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT3')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT5')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.335\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.672\n",
            "Features:  ('X2_all', 'X1_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.764\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.456\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.329\n",
            "Features:  ('X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.678\n",
            "Features:  ('X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT5')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.458\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.336\n",
            "Features:  ('X2_all', 'X1_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.694\n",
            "Features:  ('X2_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.461\n",
            "Features:  ('X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.459\n",
            "Features:  ('X2_all', 'X1_all', 'X_bursts_all', 'X_cumsum_all', 'X_CAT1', 'X_CAT2', 'X_CAT4', 'X_CAT3', 'X_CAT5')\n",
            "Accuracy: 0.461\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "# 결과를 정확도에 따라 내림차순으로 정렬\n",
        "sorted_results = sorted(results.items(), key=lambda x: x[1], reverse=True)\n",
        "\n",
        "# 상위 10개만 선택\n",
        "sorted_results = sorted_results[:10]\n",
        "\n",
        "# 피처 조합과 정확도를 분리\n",
        "feature_combinations, accuracies = zip(*sorted_results)\n",
        "\n",
        "# 피처 조합을 문자열로 변환\n",
        "feature_combinations = ['+'.join(features) for features in feature_combinations]\n",
        "\n",
        "# 막대 그래프 생성\n",
        "plt.figure(figsize=(10, 10))\n",
        "bars = plt.barh(feature_combinations, accuracies, color='skyblue')\n",
        "plt.xlabel('Accuracy')\n",
        "plt.ylabel('Feature combinations')\n",
        "plt.title('Top 10 GradientBoosting accuracies for each feature combination- unmon data included')\n",
        "\n",
        "# 결과값(result)을 막대 옆에 출력\n",
        "for bar, result in zip(bars, sorted_results):\n",
        "    plt.text(bar.get_width(), bar.get_y() + bar.get_height()/2, f'{result[1]:.4f}', va='center', ha='left')\n",
        "\n",
        "plt.gca().invert_yaxis()  # y축을 내림차순으로 정렬\n",
        "plt.show()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 872
        },
        "id": "n-64w7t64jA_",
        "outputId": "df18b1bf-d397-4fa7-f18e-dc1139a82f87"
      },
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1000x1000 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "OQ2ogkXja6rm"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}