Google Stock trend prediction with LSTM.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "rnn google stockprice.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "TPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "TnamuxPl-AgE"
      },
      "source": [
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import pandas as pd "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "j1_qRbpl-MXl"
      },
      "source": [
        "# data preprocessing\n",
        "df=pd.read_csv('/content/Google_Stock_Price_Train.csv')\n",
        "\n",
        "training_data=df.iloc[ : , 1:2].values\n",
        "# now training data variable is array of numbers"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 142
        },
        "id": "INpK5MG5eoJR",
        "outputId": "21e8114b-a8fd-47fd-f83d-51c97b3fbbb4"
      },
      "source": [
        "df.head(3)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Date</th>\n",
              "      <th>Open</th>\n",
              "      <th>High</th>\n",
              "      <th>Low</th>\n",
              "      <th>Close</th>\n",
              "      <th>Volume</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1/3/2012</td>\n",
              "      <td>325.25</td>\n",
              "      <td>332.83</td>\n",
              "      <td>324.97</td>\n",
              "      <td>663.59</td>\n",
              "      <td>7,380,500</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1/4/2012</td>\n",
              "      <td>331.27</td>\n",
              "      <td>333.87</td>\n",
              "      <td>329.08</td>\n",
              "      <td>666.45</td>\n",
              "      <td>5,749,400</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>1/5/2012</td>\n",
              "      <td>329.83</td>\n",
              "      <td>330.75</td>\n",
              "      <td>326.89</td>\n",
              "      <td>657.21</td>\n",
              "      <td>6,590,300</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       Date    Open    High     Low   Close     Volume\n",
              "0  1/3/2012  325.25  332.83  324.97  663.59  7,380,500\n",
              "1  1/4/2012  331.27  333.87  329.08  666.45  5,749,400\n",
              "2  1/5/2012  329.83  330.75  326.89  657.21  6,590,300"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6vs4mUGe-Mal"
      },
      "source": [
        "from sklearn.preprocessing import MinMaxScaler\n",
        "mm=MinMaxScaler()\n",
        "scaled_training_data=mm.fit_transform(training_data)\n",
        "# normalization is best to use when output layer has sigmoid activation"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "to8FyVDL-MdV"
      },
      "source": [
        "X_train=[]\n",
        "y_train=[]\n",
        "for i in range(60,1258):\n",
        "  X_train.append(scaled_training_data[i-60:i,0])\n",
        "  y_train.append(scaled_training_data[i,0])\n",
        "X_train,y_train=np.array(X_train),np.array(y_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IhwasVdchKIF",
        "outputId": "5f14a0f6-8c8b-4f9c-8820-6ded4e2122ce"
      },
      "source": [
        "X_train.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(1198, 60)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "l7bMf8k6kh_2"
      },
      "source": [
        "# reshaping took place as \n",
        "X_train=np.reshape(X_train,(1198,60,1))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VkSr82-ikiLI"
      },
      "source": [
        "# start rnn\n",
        "# import req libraries\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Dense\n",
        "from keras.layers import LSTM\n",
        "from keras.layers import Dropout"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DkFozrujlSeu"
      },
      "source": [
        "# initialize rnn\n",
        "rnn=Sequential()\n",
        "rnn.add(LSTM(units=50,return_sequences=True,input_shape=(60,1)))\n",
        "rnn.add(Dropout(rate=0.2))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vODsoFDAlSmT"
      },
      "source": [
        "# adding 2 nd,3 rd,4th layer of LSTM network\n",
        "rnn.add(LSTM(units=50,return_sequences=True))\n",
        "rnn.add(Dropout(rate=0.2))\n",
        "rnn.add(LSTM(units=50,return_sequences=True))\n",
        "rnn.add(Dropout(rate=0.2))\n",
        "rnn.add(LSTM(units=50))\n",
        "rnn.add(Dropout(rate=0.2))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iBR8SFAIlSt1"
      },
      "source": [
        "# add output layer\n",
        "rnn.add(Dense(units=1))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8p6Pa13akiOO"
      },
      "source": [
        "#adding optimizer and loss func\n",
        "rnn.compile(optimizer='adam',loss='mean_squared_error')\n",
        "# generally rms prop is good optimizer for recurent neuralnetworks but we can also use adam\n",
        "#generally mean square error is loss func in regression prblms"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mS1SN27cxp13",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "f5b2b848-ef25-480c-e8d3-693fad2c5eed"
      },
      "source": [
        "# now inputing train data \n",
        "rnn.fit(X_train,y_train,batch_size=32,epochs=5)\n",
        "# here i changed order shown in video epoch 1st batch_size later"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/5\n",
            "38/38 [==============================] - 10s 106ms/step - loss: 0.0964\n",
            "Epoch 2/5\n",
            "38/38 [==============================] - 4s 109ms/step - loss: 0.0068\n",
            "Epoch 3/5\n",
            "38/38 [==============================] - 4s 110ms/step - loss: 0.0058\n",
            "Epoch 4/5\n",
            "38/38 [==============================] - 4s 112ms/step - loss: 0.0052\n",
            "Epoch 5/5\n",
            "38/38 [==============================] - 4s 111ms/step - loss: 0.0052\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7f9e5f835978>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9suh5n2Exp5k"
      },
      "source": [
        "# now getting real test data\n",
        "df2=pd.read_csv('/content/Google_Stock_Price_Test.csv')\n",
        "test_data=df2.iloc[ : ,1:2].values\n",
        "# the above step convert dataframe to numpy array"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cPdDZ_2Vxp-X"
      },
      "source": [
        "# now concat dataframes train and test\n",
        "df3=pd.concat((df['Open'],df2['Open']),axis=0)\n",
        "final_inputs=df3[len(df3)-len(df2)-60 : ].values\n",
        "inputs=final_inputs.reshape(-1,1)\n",
        "inputs=mm.transform(inputs)\n",
        "# no need fit becoz we already done it initially with train data\n",
        "# now we convert to 3D array format\n",
        "X_test=[]\n",
        "for i in range(60,80):\n",
        "  X_test.append(inputs[i-60:i,0])\n",
        "X_test=np.array(X_test)\n",
        "X_test=np.reshape(X_test,(X_test.shape[0],X_test.shape[1],1))\n",
        "# now we can predict stock price and we can rescale back predictions to normal way\n",
        "estimated_stockprice=rnn.predict(X_test)\n",
        "estimated_stockprice=mm.inverse_transform(estimated_stockprice)  "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Be1TWaFa5dNT",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "outputId": "e37cec06-30b5-403c-ffda-f1c2dedb6bea"
      },
      "source": [
        "# now ploting both trnds and compare them\n",
        "plt.plot(test_data,color='green',label='real_stockvalue')\n",
        "plt.plot(estimated_stockprice,color='red',label='predicted_stockvalue')\n",
        "plt.title('google stock price estimation with rnn')\n",
        "plt.xlabel('financial_day')\n",
        "plt.ylabel('stock_value')\n",
        "plt.legend()\n",
        "plt.show()\n",
        "# the above function show will show the plot"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}