model_sign_language_translator_2.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 121,
      "metadata": {
        "id": "Fm_WS5SkanIP"
      },
      "outputs": [],
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "df = pd.read_excel(\"/content/ASL_alphabets.xlsx\", index_col=False)"
      ],
      "metadata": {
        "id": "EB0Xf3ywbAkk"
      },
      "execution_count": 122,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# https://www.geeksforgeeks.org/how-to-drop-one-or-multiple-columns-in-pandas-dataframe/\n",
        "df = df.drop([0,1],axis=1)\n",
        "print(len(df))\n",
        "print(df.head())\n",
        "# [0,1,2,3,4] -> [A,Y,U,S,H]"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TW0X0JqVbEvI",
        "outputId": "bfe050f6-7f95-47d3-ec03-d43e19dc2a96"
      },
      "execution_count": 123,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "1000\n",
            "   Unnamed: 0  class         2         3         4         5         6  \\\n",
            "0           0      0 -0.083955 -0.057687 -0.143967 -0.166604 -0.158132   \n",
            "1           1      0 -0.083900 -0.061100 -0.141032 -0.165363 -0.158884   \n",
            "2           2      0 -0.086185 -0.057912 -0.151771 -0.144644 -0.180921   \n",
            "3           3      0 -0.084918 -0.037873 -0.150861 -0.124648 -0.172609   \n",
            "4           4      0 -0.080348 -0.028174 -0.146617 -0.121466 -0.170870   \n",
            "\n",
            "          7         8         9  ...        32        33        34        35  \\\n",
            "0 -0.270313 -0.146678 -0.353363  ...  0.008286 -0.224137  0.067247 -0.265167   \n",
            "1 -0.270144 -0.152182 -0.352973  ...  0.000033 -0.222951  0.056077 -0.252391   \n",
            "2 -0.243660 -0.172599 -0.326726  ... -0.028672 -0.220826  0.027602 -0.257759   \n",
            "3 -0.227299 -0.169301 -0.307612  ... -0.027564 -0.208170  0.024663 -0.242198   \n",
            "4 -0.225190 -0.166369 -0.303471  ... -0.027546 -0.204892  0.024202 -0.238494   \n",
            "\n",
            "         36        37        38        39        40        41  \n",
            "0  0.064527 -0.321461  0.054204 -0.252515  0.051229 -0.222619  \n",
            "1  0.053019 -0.318152  0.041167 -0.252924  0.037451 -0.220153  \n",
            "2  0.024671 -0.321016  0.015170 -0.251215  0.013000 -0.224307  \n",
            "3  0.024185 -0.302602  0.018889 -0.233397  0.013695 -0.208750  \n",
            "4  0.024323 -0.299691  0.019614 -0.233342  0.013261 -0.208376  \n",
            "\n",
            "[5 rows x 42 columns]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# the df has index for some reason and we need to remove it\n",
        "# df.drop('Unnamed: 0',axis=1)\n",
        "df = df.iloc[:,1:]\n",
        "print(df.head())"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8M7-T9Jt0M9A",
        "outputId": "3f3525e5-6b33-4d2a-8e9d-30ba8046eed3"
      },
      "execution_count": 124,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "   class         2         3         4         5         6         7  \\\n",
            "0      0 -0.083955 -0.057687 -0.143967 -0.166604 -0.158132 -0.270313   \n",
            "1      0 -0.083900 -0.061100 -0.141032 -0.165363 -0.158884 -0.270144   \n",
            "2      0 -0.086185 -0.057912 -0.151771 -0.144644 -0.180921 -0.243660   \n",
            "3      0 -0.084918 -0.037873 -0.150861 -0.124648 -0.172609 -0.227299   \n",
            "4      0 -0.080348 -0.028174 -0.146617 -0.121466 -0.170870 -0.225190   \n",
            "\n",
            "          8         9        10  ...        32        33        34        35  \\\n",
            "0 -0.146678 -0.353363 -0.100944  ...  0.008286 -0.224137  0.067247 -0.265167   \n",
            "1 -0.152182 -0.352973 -0.100487  ...  0.000033 -0.222951  0.056077 -0.252391   \n",
            "2 -0.172599 -0.326726 -0.127894  ... -0.028672 -0.220826  0.027602 -0.257759   \n",
            "3 -0.169301 -0.307612 -0.124981  ... -0.027564 -0.208170  0.024663 -0.242198   \n",
            "4 -0.166369 -0.303471 -0.122225  ... -0.027546 -0.204892  0.024202 -0.238494   \n",
            "\n",
            "         36        37        38        39        40        41  \n",
            "0  0.064527 -0.321461  0.054204 -0.252515  0.051229 -0.222619  \n",
            "1  0.053019 -0.318152  0.041167 -0.252924  0.037451 -0.220153  \n",
            "2  0.024671 -0.321016  0.015170 -0.251215  0.013000 -0.224307  \n",
            "3  0.024185 -0.302602  0.018889 -0.233397  0.013695 -0.208750  \n",
            "4  0.024323 -0.299691  0.019614 -0.233342  0.013261 -0.208376  \n",
            "\n",
            "[5 rows x 41 columns]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.decomposition import PCA\n",
        "# https://scikit-learn.org/stable/modules/generated/sklearn.decomposition.PCA.html\n",
        "\n",
        "X = df\n",
        "X = X.drop([\"class\"],axis=1)#.reset_index(drop=True)\n",
        "print(X.head())\n",
        "\n",
        "X.columns = X.columns.astype(str)\n",
        "pca = PCA()\n",
        "pca_data = pca.fit_transform(X)\n",
        "pca_data"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ofIcrUaJbIZ4",
        "outputId": "41f32062-6235-47de-9f39-e558833cea93"
      },
      "execution_count": 125,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "         2         3         4         5         6         7         8   \\\n",
            "0 -0.083955 -0.057687 -0.143967 -0.166604 -0.158132 -0.270313 -0.146678   \n",
            "1 -0.083900 -0.061100 -0.141032 -0.165363 -0.158884 -0.270144 -0.152182   \n",
            "2 -0.086185 -0.057912 -0.151771 -0.144644 -0.180921 -0.243660 -0.172599   \n",
            "3 -0.084918 -0.037873 -0.150861 -0.124648 -0.172609 -0.227299 -0.169301   \n",
            "4 -0.080348 -0.028174 -0.146617 -0.121466 -0.170870 -0.225190 -0.166369   \n",
            "\n",
            "         9         10        11  ...        32        33        34        35  \\\n",
            "0 -0.353363 -0.100944 -0.278426  ...  0.008286 -0.224137  0.067247 -0.265167   \n",
            "1 -0.352973 -0.100487 -0.272005  ...  0.000033 -0.222951  0.056077 -0.252391   \n",
            "2 -0.326726 -0.127894 -0.254771  ... -0.028672 -0.220826  0.027602 -0.257759   \n",
            "3 -0.307612 -0.124981 -0.237485  ... -0.027564 -0.208170  0.024663 -0.242198   \n",
            "4 -0.303471 -0.122225 -0.230068  ... -0.027546 -0.204892  0.024202 -0.238494   \n",
            "\n",
            "         36        37        38        39        40        41  \n",
            "0  0.064527 -0.321461  0.054204 -0.252515  0.051229 -0.222619  \n",
            "1  0.053019 -0.318152  0.041167 -0.252924  0.037451 -0.220153  \n",
            "2  0.024671 -0.321016  0.015170 -0.251215  0.013000 -0.224307  \n",
            "3  0.024185 -0.302602  0.018889 -0.233397  0.013695 -0.208750  \n",
            "4  0.024323 -0.299691  0.019614 -0.233342  0.013261 -0.208376  \n",
            "\n",
            "[5 rows x 40 columns]\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[-2.38564165e-01, -9.54846244e-02, -1.56708177e-01, ...,\n",
              "         5.88927302e-06,  1.97797189e-05,  1.80942234e-04],\n",
              "       [-2.53655597e-01, -1.23696752e-01, -1.44804715e-01, ...,\n",
              "        -1.06408734e-04, -2.27935160e-04,  5.34639706e-05],\n",
              "       [-2.89854120e-01, -2.26712335e-01, -1.55195661e-01, ...,\n",
              "        -5.96195058e-04,  4.37693176e-04,  7.33746828e-05],\n",
              "       ...,\n",
              "       [ 3.97314172e-01,  1.88813707e-03, -8.30351951e-02, ...,\n",
              "        -6.30112411e-05,  1.66207777e-04,  1.61636689e-05],\n",
              "       [ 3.98094875e-01,  4.10567675e-03, -8.82119557e-02, ...,\n",
              "         1.94769958e-04,  2.88443871e-04,  1.50121277e-06],\n",
              "       [ 3.97413671e-01,  1.12522706e-02, -7.30052219e-02, ...,\n",
              "        -1.37898694e-04,  6.27701660e-05,  8.21181979e-05]])"
            ]
          },
          "metadata": {},
          "execution_count": 125
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(pca_data)\n",
        "\n",
        "print(len(pca.explained_variance_ratio_))\n",
        "# print(pca.explained_variance_ratio_)\n",
        "\n",
        "plt.bar(range(1,len(pca.explained_variance_)+1),height=pca.explained_variance_ratio_)\n",
        "plt.xlabel('PCA Feature')\n",
        "plt.ylabel('Explained variance')\n",
        "plt.title('Feature Explained Variance')\n",
        "plt.show()\n",
        "\n",
        "# https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.bar.html\n",
        "\n",
        "# As we can see , even 1 dimension in PCA plot can account for all the variance in the data\n",
        "\n",
        "df_pca = pd.DataFrame(data=pca_data[:,:2], columns=['PC1','PC2'])\n",
        "df_pca['class'] = df['class']\n",
        "\n",
        "print(df_pca.head())\n",
        "\n",
        "# # Plot using sns.kdeplot\n",
        "# sns.kdeplot(data=df_pca, x='PC1', hue='class')\n",
        "sns.scatterplot(data=df_pca,x=\"PC1\",y=\"PC2\",hue=\"class\",palette=\"deep\")\n",
        "\n",
        "# Show the plot\n",
        "plt.show()\n",
        "\n",
        "# As we can see from the kde plot of the PCA'ed data, the classes should be linearly separable"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "DwIxo1gbch1h",
        "outputId": "85fc58b8-f30a-41dc-f110-d6d17f6f7fef"
      },
      "execution_count": 126,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[-2.38564165e-01 -9.54846244e-02 -1.56708177e-01 ...  5.88927302e-06\n",
            "   1.97797189e-05  1.80942234e-04]\n",
            " [-2.53655597e-01 -1.23696752e-01 -1.44804715e-01 ... -1.06408734e-04\n",
            "  -2.27935160e-04  5.34639706e-05]\n",
            " [-2.89854120e-01 -2.26712335e-01 -1.55195661e-01 ... -5.96195058e-04\n",
            "   4.37693176e-04  7.33746828e-05]\n",
            " ...\n",
            " [ 3.97314172e-01  1.88813707e-03 -8.30351951e-02 ... -6.30112411e-05\n",
            "   1.66207777e-04  1.61636689e-05]\n",
            " [ 3.98094875e-01  4.10567675e-03 -8.82119557e-02 ...  1.94769958e-04\n",
            "   2.88443871e-04  1.50121277e-06]\n",
            " [ 3.97413671e-01  1.12522706e-02 -7.30052219e-02 ... -1.37898694e-04\n",
            "   6.27701660e-05  8.21181979e-05]]\n",
            "40\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "        PC1       PC2  class\n",
            "0 -0.238564 -0.095485      0\n",
            "1 -0.253656 -0.123697      0\n",
            "2 -0.289854 -0.226712      0\n",
            "3 -0.318213 -0.230614      0\n",
            "4 -0.321691 -0.230562      0\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.linear_model import LogisticRegression\n",
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "y = df['class']\n",
        "# y = y.iloc[:,1:]\n",
        "print(y.head())\n",
        "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42) #,shuffle=True\n",
        "\n",
        "# X_train = X_train.reset_index(drop=True)\n",
        "# y_train = y_train.reset_index(drop=True)\n",
        "# X_test = X_test.reset_index(drop=True)\n",
        "# y_test = y_test.reset_index(drop=True)\n",
        "\n",
        "print(y_test.head())\n",
        "\n",
        "clf = LogisticRegression(penalty=None,random_state=0, max_iter=4000).fit(X_train, y_train)\n",
        "\n",
        "print(clf.score(X_test, y_test))\n",
        "\n",
        "# https://scikit-learn.org/stable/modules/generated/sklearn.model_selection.train_test_split.html\n",
        "# https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LogisticRegression.html\n",
        "# As per the docs, logistic regression will automatically adjust to multi-class problem\n",
        "# Also regularization is automatically done, so I had to remove it\n",
        "# Bcoz in a problem like this, the variable distribution will not change much so it is actually good to overfit\n",
        "\n",
        "# We got 100% accuracy\n",
        "# logistic regression is working so well that I can be sure that data is linearly separable"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KYYIFfMedQye",
        "outputId": "0aed8161-3de4-400c-d1d6-49d9f9df3517"
      },
      "execution_count": 127,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0    0\n",
            "1    0\n",
            "2    0\n",
            "3    0\n",
            "4    0\n",
            "Name: class, dtype: int64\n",
            "521    2\n",
            "737    3\n",
            "740    3\n",
            "660    3\n",
            "411    2\n",
            "Name: class, dtype: int64\n",
            "1.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# The PCA plot was suggestive that the data may be gaussian\n",
        "# (which I think it should be since hand position distrbution is same for each alphabet)\n",
        "\n",
        "# https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.mean.html\n",
        "\n",
        "# to see a class mean\n",
        "# print(df.loc[df['class']==0].mean())\n",
        "\n",
        "for i in range(0,5):\n",
        "  print(df.loc[df['class']==i].mean().tolist())"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "dx2LoZ4zoif2",
        "outputId": "c3f5e7ae-3133-4c0a-8497-6dd2c5aa379b"
      },
      "execution_count": 128,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0.0, -0.07474212601780891, -0.0431456908583641, -0.12400060549378394, -0.13857888400554658, -0.1391511856019497, -0.23397291466593742, -0.13576779872179032, -0.30704155467450617, -0.09403851538896561, -0.22954849600791932, -0.09825520560145379, -0.30056981727480886, -0.09018444091081619, -0.2255919347703457, -0.08281404092907906, -0.190639261752367, -0.051667416989803316, -0.23708705291152, -0.05586740642786026, -0.30377587005496026, -0.053816703110933305, -0.2115647216141224, -0.050149403512477875, -0.1910230451822281, -0.007575704455375671, -0.234019705504179, -0.010493827611207962, -0.30251623436808583, -0.015133990794420243, -0.21544239580631255, -0.013848295360803604, -0.18879091516137123, 0.037870592176914214, -0.22238519221544265, 0.031362849175930026, -0.2804639020562172, 0.02113809287548065, -0.221747654825449, 0.019811767637729644, -0.19482029259204864]\n",
            "[1.0, -0.06988532483577728, -0.0695784392952919, -0.10177638247609139, -0.19462273985147477, -0.11659591041505336, -0.3030320896208286, -0.15474548250436782, -0.36645681455731394, -0.004105231463909149, -0.2829523867368698, 0.036329239308834076, -0.3468018507957458, 0.030634361207485198, -0.2684305553138256, 0.01722517728805542, -0.2042035061120987, 0.04979571402072906, -0.2681789319217205, 0.08853191360831261, -0.3404541800916195, 0.06811451941728591, -0.2481471787393093, 0.04628616988658905, -0.1824572503566742, 0.09501648277044296, -0.2425807647407055, 0.1318397991359234, -0.3089808204770088, 0.10687661200761794, -0.2353213720023632, 0.08085451319813729, -0.18027152732014656, 0.1307143761217594, -0.20993620842695238, 0.18012661889195442, -0.2894254192709923, 0.2127226234972477, -0.3329757168889046, 0.23908768460154534, -0.38225056797266005]\n",
            "[2.0, -0.05387806296348572, -0.06572277277708054, -0.06608499735593795, -0.17100832223892212, -0.02166170209646225, -0.24699612081050873, 0.02443588972091675, -0.285488478243351, -0.06190463602542877, -0.2832739716768265, -0.0622423106431961, -0.4015065771341324, -0.05827069818973541, -0.47625113859772683, -0.05342988193035126, -0.5393039385974407, -0.016729598045349122, -0.2824899211525917, -0.020153537690639496, -0.4099024772644043, -0.01979645848274231, -0.490636382997036, -0.01495338350534439, -0.5574186624586582, 0.027198221385478973, -0.25381936371326447, 0.024727240800857545, -0.3095869544148445, 0.0024276039004325865, -0.2395319539308548, -0.009403225779533387, -0.1847696489095688, 0.06422279238700866, -0.2079317733645439, 0.0533997717499733, -0.25130320370197295, 0.02604701280593872, -0.19671969443559648, 0.00931252121925354, -0.14750671058893203]\n",
            "[3.0, -0.05908229023218155, -0.04738905340433121, -0.092749462723732, -0.1309749135375023, -0.07338976323604583, -0.20622837647795678, -0.03293219953775406, -0.23890226148068905, -0.09024297058582306, -0.20119220688939093, -0.10233275532722473, -0.2692642351984978, -0.09197820782661438, -0.21878927387297153, -0.08145415455102921, -0.1853222019970417, -0.04857613921165466, -0.21287739902734756, -0.05513614356517792, -0.28405413940548896, -0.05094160109758377, -0.22973417393863202, -0.04645854741334915, -0.20212645471096038, -0.006020570695400238, -0.21263081327080727, -0.009427164196968079, -0.2767147096991539, -0.012579258680343628, -0.226794221252203, -0.01283116579055786, -0.2021825958788395, 0.03723597884178162, -0.20230237364768983, 0.03133391708135605, -0.2533623301237822, 0.023534131348133088, -0.21523480728268624, 0.021235769391059877, -0.1883329454064369]\n",
            "[4.0, -0.04359712138772011, -0.07584065347909927, -0.0311322683095932, -0.19092928007245064, 0.01864173501729965, -0.26165098398923875, 0.058979621976614, -0.3046345120668411, 0.006719292253255844, -0.25402470991015436, 0.03618919134140015, -0.3616499634087086, 0.060635415017604825, -0.42780080109834673, 0.0845454102754593, -0.48231582999229433, 0.04629799157381058, -0.23365405425429345, 0.07668589368462563, -0.35095252946019173, 0.09992716148495674, -0.4241746015101671, 0.12451086044311524, -0.4832889302074909, 0.08133260652422905, -0.19347498714923858, 0.08476168066263198, -0.2619226694107056, 0.050174227207899096, -0.1942438106238842, 0.030121835619211196, -0.13752673745155333, 0.11013815864920616, -0.14287336751818658, 0.1031722842156887, -0.1977728572487831, 0.07144496992230415, -0.1498911990225315, 0.05273885428905487, -0.10614281207323074]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.mixture import GaussianMixture\n",
        "gm = GaussianMixture(n_components=5, random_state=0).fit(X_train)\n",
        "print(gm.means_)\n",
        "print(gm.covariances_)\n",
        "\n",
        "print()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "n79DKQAnqjBX",
        "outputId": "bc51d120-cace-4d47-dba5-923706cd3ad1"
      },
      "execution_count": 129,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[-5.92778676e-02 -4.68054627e-02 -9.34450293e-02 -1.30421090e-01\n",
            "  -7.50045155e-02 -2.06132192e-01 -3.53100054e-02 -2.39613315e-01\n",
            "  -9.03532894e-02 -2.01680599e-01 -1.01634416e-01 -2.68574840e-01\n",
            "  -9.12832917e-02 -2.17065395e-01 -8.13257245e-02 -1.84463197e-01\n",
            "  -4.85514903e-02 -2.13418656e-01 -5.44871359e-02 -2.82896177e-01\n",
            "  -5.03326414e-02 -2.26872549e-01 -4.64267360e-02 -2.00509908e-01\n",
            "  -5.84724261e-03 -2.13155669e-01 -8.66648541e-03 -2.75857503e-01\n",
            "  -1.18495583e-02 -2.24452350e-01 -1.25549537e-02 -2.00656875e-01\n",
            "   3.74421249e-02 -2.02834418e-01  3.21731780e-02 -2.52794494e-01\n",
            "   2.43499783e-02 -2.13821550e-01  2.16884493e-02 -1.87629527e-01]\n",
            " [-4.89205544e-02 -7.09390085e-02 -4.92248704e-02 -1.81215074e-01\n",
            "  -2.32852365e-03 -2.54872189e-01  4.09936837e-02 -2.95552428e-01\n",
            "  -2.83371034e-02 -2.69384175e-01 -1.42014030e-02 -3.82502562e-01\n",
            "  -2.93333329e-04 -4.53039283e-01  1.38207085e-02 -5.11898168e-01\n",
            "   1.41099403e-02 -2.58979814e-01  2.71131839e-02 -3.81593764e-01\n",
            "   3.85333513e-02 -4.58601449e-01  5.29655699e-02 -5.21578294e-01\n",
            "   5.37359736e-02 -2.24596465e-01  5.39364756e-02 -2.86729999e-01\n",
            "   2.55255776e-02 -2.17860139e-01  9.70551213e-03 -1.62111902e-01\n",
            "   8.67845772e-02 -1.76249102e-01  7.76686337e-02 -2.25476382e-01\n",
            "   4.80626477e-02 -1.74335770e-01  3.03358280e-02 -1.27834840e-01]\n",
            " [-7.47998728e-02 -4.37797871e-02 -1.23963460e-01 -1.39397565e-01\n",
            "  -1.39291434e-01 -2.34781712e-01 -1.36311416e-01 -3.08212332e-01\n",
            "  -9.39509553e-02 -2.29165097e-01 -9.88923457e-02 -3.01720044e-01\n",
            "  -9.07802799e-02 -2.26885324e-01 -8.28896076e-02 -1.91048337e-01\n",
            "  -5.18108561e-02 -2.36519913e-01 -5.67510002e-02 -3.04856052e-01\n",
            "  -5.45444154e-02 -2.12867562e-01 -5.02233571e-02 -1.91400514e-01\n",
            "  -7.89839743e-03 -2.33515049e-01 -1.15971071e-02 -3.03669928e-01\n",
            "  -1.60919953e-02 -2.16667138e-01 -1.41893216e-02 -1.89267683e-01\n",
            "   3.75225797e-02 -2.22019183e-01  3.02045425e-02 -2.81605217e-01\n",
            "   2.00593467e-02 -2.22980239e-01  1.93341005e-02 -1.95391541e-01]\n",
            " [-7.44657179e-02 -6.37362145e-02 -1.14848050e-01 -1.88317311e-01\n",
            "  -1.36550696e-01 -2.97801620e-01 -1.79183659e-01 -3.56576116e-01\n",
            "  -1.73240410e-02 -2.87584536e-01  2.40113594e-02 -3.56535977e-01\n",
            "   2.36782736e-02 -2.76349257e-01  1.24354133e-02 -2.11534018e-01\n",
            "   3.90935595e-02 -2.77356123e-01  7.87432930e-02 -3.54782418e-01\n",
            "   6.37192003e-02 -2.59049655e-01  4.34955422e-02 -1.92680806e-01\n",
            "   8.66283202e-02 -2.55661385e-01  1.23261303e-01 -3.26712976e-01\n",
            "   1.03190284e-01 -2.49529514e-01  7.93051143e-02 -1.93271369e-01\n",
            "   1.23928736e-01 -2.25468228e-01  1.71118623e-01 -3.10094770e-01\n",
            "   2.02944480e-01 -3.55000795e-01  2.27389443e-01 -4.06578158e-01]\n",
            " [-6.64568618e-02 -7.41261350e-02 -9.16513787e-02 -1.99619523e-01\n",
            "  -1.01183601e-01 -3.07334672e-01 -1.36105512e-01 -3.74125235e-01\n",
            "   5.77627719e-03 -2.79475489e-01  4.50345813e-02 -3.39336030e-01\n",
            "   3.52144521e-02 -2.62209673e-01  2.01299639e-02 -1.98323907e-01\n",
            "   5.78689846e-02 -2.61390488e-01  9.54520030e-02 -3.29651710e-01\n",
            "   7.08426654e-02 -2.39895013e-01  4.77671877e-02 -1.74649014e-01\n",
            "   1.01377752e-01 -2.32941062e-01  1.37945728e-01 -2.95613914e-01\n",
            "   1.09063699e-01 -2.24532549e-01  8.12721369e-02 -1.70241332e-01\n",
            "   1.35896212e-01 -1.98596242e-01  1.86854210e-01 -2.74232031e-01\n",
            "   2.19916652e-01 -3.16972906e-01  2.47587641e-01 -3.64740342e-01]]\n",
            "[[[ 1.68498029e-05  6.90010612e-06  2.21177794e-05 ...  2.37743132e-05\n",
            "    1.02406166e-05  1.47848760e-05]\n",
            "  [ 6.90010612e-06  4.33921937e-05 -9.67197143e-06 ...  6.33314430e-05\n",
            "    1.66503646e-06  3.88358867e-05]\n",
            "  [ 2.21177794e-05 -9.67197143e-06  5.64228562e-05 ... -2.32208798e-05\n",
            "    2.81070506e-05 -1.78450340e-05]\n",
            "  ...\n",
            "  [ 2.37743132e-05  6.33314430e-05 -2.32208798e-05 ...  2.13656527e-04\n",
            "    1.58103794e-06  1.37338345e-04]\n",
            "  [ 1.02406166e-05  1.66503646e-06  2.81070506e-05 ...  1.58103794e-06\n",
            "    4.74274081e-05 -1.15908064e-05]\n",
            "  [ 1.47848760e-05  3.88358867e-05 -1.78450340e-05 ...  1.37338345e-04\n",
            "   -1.15908064e-05  9.80989962e-05]]\n",
            "\n",
            " [[ 6.50827331e-05 -2.86828525e-05  1.78585299e-04 ...  2.20319376e-04\n",
            "    1.57783621e-04  1.86248021e-04]\n",
            "  [-2.86828525e-05  1.47791811e-04 -1.28353743e-04 ...  8.76625501e-05\n",
            "   -1.47388868e-04  9.36667967e-05]\n",
            "  [ 1.78585299e-04 -1.28353743e-04  5.41898536e-04 ...  5.98344664e-04\n",
            "    5.06840966e-04  5.07510760e-04]\n",
            "  ...\n",
            "  [ 2.20319376e-04  8.76625501e-05  5.98344664e-04 ...  1.37583206e-03\n",
            "    5.31728122e-04  1.22371224e-03]\n",
            "  [ 1.57783621e-04 -1.47388868e-04  5.06840966e-04 ...  5.31728122e-04\n",
            "    5.86991899e-04  4.59862665e-04]\n",
            "  [ 1.86248021e-04  9.36667967e-05  5.07510760e-04 ...  1.22371224e-03\n",
            "    4.59862665e-04  1.10098622e-03]]\n",
            "\n",
            " [[ 8.64940498e-05 -8.21510847e-06  1.37993788e-04 ...  1.51829349e-04\n",
            "    5.92433592e-05  8.76334138e-05]\n",
            "  [-8.21510847e-06  6.17066813e-05 -4.67849825e-05 ... -9.68417962e-06\n",
            "   -2.44506807e-05 -1.31720179e-05]\n",
            "  [ 1.37993788e-04 -4.67849825e-05  2.52920775e-04 ...  2.51751463e-04\n",
            "    9.57826516e-05  1.60423872e-04]\n",
            "  ...\n",
            "  [ 1.51829349e-04 -9.68417962e-06  2.51751463e-04 ...  3.60703976e-04\n",
            "    9.67122119e-05  2.29660542e-04]\n",
            "  [ 5.92433592e-05 -2.44506807e-05  9.57826516e-05 ...  9.67122119e-05\n",
            "    9.12888578e-05  4.33556449e-05]\n",
            "  [ 8.76334138e-05 -1.31720179e-05  1.60423872e-04 ...  2.29660542e-04\n",
            "    4.33556449e-05  1.70857363e-04]]\n",
            "\n",
            " [[ 1.29518166e-05 -7.96547545e-06  1.79804745e-05 ...  3.34094271e-05\n",
            "    1.15565293e-05  4.16709733e-05]\n",
            "  [-7.96547545e-06  1.32998990e-05 -1.03542733e-05 ... -1.89942679e-05\n",
            "   -1.06663703e-05 -2.50014184e-05]\n",
            "  [ 1.79804745e-05 -1.03542733e-05  3.57872169e-05 ...  6.68474162e-05\n",
            "   -1.78135773e-05  8.00084203e-05]\n",
            "  ...\n",
            "  [ 3.34094271e-05 -1.89942679e-05  6.68474162e-05 ...  1.44655097e-04\n",
            "   -3.94245858e-05  1.72113165e-04]\n",
            "  [ 1.15565293e-05 -1.06663703e-05 -1.78135773e-05 ... -3.94245858e-05\n",
            "    7.09694198e-04 -5.44460505e-05]\n",
            "  [ 4.16709733e-05 -2.50014184e-05  8.00084203e-05 ...  1.72113165e-04\n",
            "   -5.44460505e-05  2.12521949e-04]]\n",
            "\n",
            " [[ 1.47173038e-05 -2.17686128e-05  4.02118962e-05 ...  8.41990497e-06\n",
            "    3.44954317e-06  4.69785441e-06]\n",
            "  [-2.17686128e-05  7.92567300e-05 -8.84966541e-05 ...  1.71987403e-05\n",
            "   -5.55398768e-05  5.32151872e-05]\n",
            "  [ 4.02118962e-05 -8.84966541e-05  1.38975152e-04 ...  8.14003796e-06\n",
            "    3.82439097e-05 -2.36044547e-05]\n",
            "  ...\n",
            "  [ 8.41990497e-06  1.71987403e-05  8.14003796e-06 ...  1.41180066e-04\n",
            "   -8.27489762e-05  2.15971981e-04]\n",
            "  [ 3.44954317e-06 -5.55398768e-05  3.82439097e-05 ... -8.27489762e-05\n",
            "    2.36123681e-04 -1.77647250e-04]\n",
            "  [ 4.69785441e-06  5.32151872e-05 -2.36044547e-05 ...  2.15971981e-04\n",
            "   -1.77647250e-04  3.57310811e-04]]]\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# oh right , since there was no y_train in the GMM training so the indices could be in any order\n",
        "# and whenever we retrain it, it could be in any fucking order, so need to make a general algorithm to detect alphabets\n",
        "\n",
        "# My idea - if it is predicting with probablity > 0.99 then we can assign it to the one whose mean it is closest to\n",
        "\n",
        "cluster_means = []\n",
        "for i in range(0,5):\n",
        "  cluster_means.append(df.loc[df['class']==i].mean().drop('class').values)\n",
        "\n",
        "cluster_means = np.array(cluster_means)\n",
        "\n",
        "\n",
        "# print(np.max(gm.predict_proba(X_test),axis=1))\n",
        "\n",
        "# if(np.max(gm.predict_proba(X_test),axis=1)>0.99):\n",
        "#   for i in range(0,len(cluster_means)):\n",
        "\n",
        "# print(y_test.tolist())"
      ],
      "metadata": {
        "id": "7Hj4T2FQz_KC"
      },
      "execution_count": 130,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Whole heartedly thanking chatgpt for the code in this cell,\n",
        "# done perfectly as per my description without any mistakes in first time\n",
        "\n",
        "from sklearn.metrics import accuracy_score\n",
        "\n",
        "# Predict probabilities for the test set\n",
        "probabilities = gm.predict_proba(X_test)\n",
        "\n",
        "# Initialize an array to store the predicted cluster assignments\n",
        "predicted_clusters = np.empty(X_test.shape[0])\n",
        "\n",
        "# Iterate over each test sample\n",
        "for idx, sample in enumerate(X_test.values):\n",
        "    if np.max(probabilities[idx]) > 0.99:\n",
        "        # Find the closest cluster mean by Euclidean distance\n",
        "        distances = np.linalg.norm(cluster_means - sample, axis=1)\n",
        "        closest_cluster = np.argmin(distances)\n",
        "        predicted_clusters[idx] = closest_cluster\n",
        "    else:\n",
        "        # Assign -1 for samples that do not meet the threshold\n",
        "        predicted_clusters[idx] = -1\n",
        "\n",
        "# Create a DataFrame for y_test and predicted_clusters to align indices\n",
        "y_test = y_test.reset_index(drop=True)\n",
        "predicted_clusters = pd.Series(predicted_clusters)\n",
        "\n",
        "# Filter out the samples that were not assigned a cluster (-1)\n",
        "filtered_predictions = predicted_clusters[predicted_clusters != -1]\n",
        "filtered_y_test = y_test[predicted_clusters != -1]\n",
        "\n",
        "# Calculate accuracy\n",
        "accuracy = accuracy_score(filtered_y_test, filtered_predictions)\n",
        "print(f\"Accuracy: {accuracy}\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "WLIYlZAB6cML",
        "outputId": "2fef2f2d-38ac-413e-d47b-3964779bbfe6"
      },
      "execution_count": 131,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Accuracy: 0.964824120603015\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(predicted_clusters)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "WnCNxNt3nbAW",
        "outputId": "84245aa2-7d0a-470a-b7d6-472de2a3fd1b"
      },
      "execution_count": 132,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0      2.0\n",
            "1      3.0\n",
            "2      3.0\n",
            "3      3.0\n",
            "4      2.0\n",
            "      ... \n",
            "195    2.0\n",
            "196    1.0\n",
            "197    1.0\n",
            "198    3.0\n",
            "199    0.0\n",
            "Length: 200, dtype: float64\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# pip install joblib"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "dqxXHm1cpzIs",
        "outputId": "4e0b5f71-6993-454f-894d-3a33d6378f64"
      },
      "execution_count": 134,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Requirement already satisfied: joblib in /usr/local/lib/python3.10/dist-packages (1.4.2)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import joblib\n",
        "\n",
        "# Training the final GMM model on the whole dataset\n",
        "gm = GaussianMixture(n_components=5, random_state=0).fit(X)\n",
        "\n",
        "# Save the model to a file\n",
        "joblib.dump(gm, 'gmm_model.pkl')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "CH9PQYaxnnnp",
        "outputId": "a71f07d3-5cc4-479c-9e7a-7df3ad7d6bd6"
      },
      "execution_count": 136,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['gmm_model.pkl']"
            ]
          },
          "metadata": {},
          "execution_count": 136
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Load the model from the file\n",
        "gm_loaded = joblib.load('gmm_model.pkl')\n"
      ],
      "metadata": {
        "id": "NXUmc7M1px-r"
      },
      "execution_count": 138,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# print(gm_loaded.means_)\n",
        "# print()\n",
        "# print(cluster_means)\n",
        "\n",
        "# Map each GMM component to the closest cluster mean\n",
        "gmm_to_cluster_map = []\n",
        "for gmm_mean in gm_loaded.means_:\n",
        "    distances = np.linalg.norm(cluster_means - gmm_mean, axis=1)\n",
        "    closest_cluster = np.argmin(distances)\n",
        "    gmm_to_cluster_map.append(closest_cluster)\n",
        "\n",
        "print(\"GMM to Cluster Map:\", gmm_to_cluster_map)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YtCu733mq_m_",
        "outputId": "0bf5bda6-0060-4f75-f4a7-cfb993af97f8"
      },
      "execution_count": 141,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "GMM to Cluster Map: [0, 4, 1, 2, 2]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "Agnw8OCbrHqk"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}