NeuralNetwork.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "NN for SM.ipynb",
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "id": "jqS29HoOuMQe",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "6df2988f-8945-4cf9-a17f-089d3b0694cb"
      },
      "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",
      "source": [
        "import numpy as np  \n",
        "import pandas as pd  \n",
        "from matplotlib import pyplot as plt \n",
        "import torch \n",
        "import torch.nn as nn\n",
        "import tensorflow as tf\n",
        "import time\n",
        "\n",
        "\n",
        "train_data_path ='/content/drive/My Drive/train.csv'\n",
        "train = pd.read_csv(train_data_path)\n",
        "\n",
        "num_of_train_data = train.shape[0]\n",
        "\n",
        "test_data_path ='/content/drive/My Drive/test.csv'\n",
        "test = pd.read_csv(test_data_path)\n",
        "\n",
        "num_of_test_data = test.shape[0]\n",
        "\n",
        "target = train[\"Capacity(State of Charge)\"] # capacity\n",
        "train.drop([\"Capacity(State of Charge)\"],axis = 1 , inplace = True) # drop target column\n",
        "\n",
        "# combine train and test\n",
        "val = test[\"Capacity(State of Charge)\"] # capacity\n",
        "test.drop([\"Capacity(State of Charge)\"],axis = 1 , inplace = True) # drop target column\n",
        "combined = train.append(test)\n",
        "combined.reset_index(inplace=True)\n",
        "combined.drop(['index'], inplace=True, axis=1)\n",
        "\n",
        "# train dataset features\n",
        "train_features = torch.tensor(combined[:num_of_train_data].values, dtype=torch.float)\n",
        "# target\n",
        "train_labels = torch.tensor(target.values, dtype=torch.float).view(-1, 1)\n",
        "\n",
        "#val_features = torch.tensor(combined[num_of_train_data:num_of_train_data+25].values, dtype=torch.float)\n",
        "#val_labels = torch.tensor(val[0:25].values, dtype=torch.float).view(-1, 1)\n",
        "\n",
        "\n",
        "test_features = torch.tensor(combined[num_of_train_data:].values, dtype=torch.float)\n",
        "test_true_value = torch.tensor(val.values, dtype=torch.float).view(-1, 1)\n",
        "\n",
        "\n",
        "print(\"train data size: \", train_features.shape)\n",
        "print(\"train label data size: \", train_labels.shape)\n",
        "#print(\"valid data size: \", val_features.shape)\n",
        "#print(\"valid label data size: \", val_labels.shape)\n",
        "print(\"test data size: \", test_features.shape)\n",
        "print(\"test label data size: \", test_true_value.shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "L050nZTqv3o0",
        "outputId": "04cfcc42-fb31-432c-a3af-40bd2755afa2"
      },
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "train data size:  torch.Size([118, 6])\n",
            "train label data size:  torch.Size([118, 1])\n",
            "test data size:  torch.Size([50, 6])\n",
            "test label data size:  torch.Size([50, 1])\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        
        "class Net(nn.Module):\n",
        "  \n",
        "    def __init__(self, features):\n",
        "        super(Net, self).__init__()\n",
        "        \n",
        "        self.linear_relu1 = nn.Linear(features, 128)\n",
        "        self.linear_relu2 = nn.Linear(128, 256)\n",
        "        self.linear_relu3 = nn.Linear(256, 256)\n",
        "        self.linear_relu4 = nn.Linear(256, 256)\n",
        "        self.linear5 = nn.Linear(256, 1)\n",
        "        \n",
        "    def forward(self, x):\n",
        "        \n",
        "        y_pred = self.linear_relu1(x)\n",
        "        y_pred = nn.functional.relu(y_pred)\n",
        "\n",
        "        y_pred = self.linear_relu2(y_pred)\n",
        "        y_pred = nn.functional.relu(y_pred)\n",
        "\n",
        "        y_pred = self.linear_relu3(y_pred)\n",
        "        y_pred = nn.functional.relu(y_pred)\n",
        "\n",
        "        y_pred = self.linear_relu4(y_pred)\n",
        "        y_pred = nn.functional.relu(y_pred)\n",
        "\n",
        "        y_pred = self.linear5(y_pred)\n",
        "        return y_pred\n",
        "\n",
        "model = Net(features=train_features.shape[1])\n",
        "\n",
        "# loss function\n",
        "criterion = nn.MSELoss(reduction='mean')\n",
        "\n",
        "optimizer = torch.optim.Adam(model.parameters(), lr=1e-4)\n"
      ],
      "metadata": {
        "id": "p1TbMo9ozU0t"
      },
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# train model\n",
        "epoch = 1000\n",
        "train_losses = []\n",
        "val_losses = []\n",
        "s_time = time.time()\n",
        "for t in range(epoch):\n",
        "    model.train()\n",
        "    y_pred = model(train_features)\n",
        "    \n",
        "    tr_loss = criterion(y_pred, train_labels)\n",
        "    train_losses.append(tr_loss.item())\n",
        "    \n",
        "    if torch.isnan(tr_loss):\n",
        "        break\n",
        "    \n",
        "    optimizer.zero_grad()\n",
        "    \n",
        "    # backward\n",
        "    tr_loss.backward()\n",
        "    optimizer.step()\n",
        "    \n",
        "    \"\"\"\n",
        "    model.eval()\n",
        "    with torch.no_grad():\n",
        "        val_pred = model(val_features)\n",
        "        val_loss = criterion(val_pred, val_labels)\n",
        "        # print(t, loss.item())\n",
        "        val_losses.append(val_loss.item())\n",
        "    \"\"\"\n",
        "\n",
        "    if t % 100 == 0:  \n",
        "        print(f\"Epoch: {t}, Train loss :{tr_loss:.3f}\") #, Val loss :{val_loss:.3f}\n",
        "c_time = time.time()\n",
        "ss_time = c_time - s_time \n",
        "print('%.3f秒'%ss_time)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BBLWbE5M0aJw",
        "outputId": "c8542c89-b811-43b4-f3f3-2174e7273afd"
      },
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch: 0, Train loss :2.556\n",
            "Epoch: 100, Train loss :0.049\n",
            "Epoch: 200, Train loss :0.021\n",
            "Epoch: 300, Train loss :0.014\n",
            "Epoch: 400, Train loss :0.010\n",
            "Epoch: 500, Train loss :0.005\n",
            "Epoch: 600, Train loss :0.002\n",
            "Epoch: 700, Train loss :0.001\n",
            "Epoch: 800, Train loss :0.000\n",
            "Epoch: 900, Train loss :0.000\n",
            "4.439秒\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.plot(np.arange(1000), train_losses)  \n",
        "#plt.plot(np.arange(1000), val_losses)\n",
        "plt.show()  \n",
        "\n",
        "plt.plot(np.arange(num_of_train_data), train_labels,\"-\",label = 'train') \n",
        "plt.plot(np.arange(num_of_train_data), y_pred.detach().numpy(),\".\",label = 'pred')  \n",
        "plt.legend() "
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 531
        },
        "id": "wv5kEXzz0lg9",
        "outputId": "89b0d89e-9ccd-4ad0-83ad-e6c9ad6be017"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7fdfa7e67610>"
            ]
          },
          "metadata": {},
          "execution_count": 7
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "predictions = model(test_features).detach().numpy()\n",
        "plt.plot(np.arange(num_of_test_data), test_true_value,\"-\",label = 'test')  \n",
        "plt.plot(np.arange(num_of_test_data), predictions,\".\",label = 'pred') \n",
        "plt.legend()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "id": "kdTgIQ2R3jr2",
        "outputId": "53357b98-e62a-4a2d-9dea-64f3da379d2e"
      },
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7fdfa7d76ed0>"
            ]
          },
          "metadata": {},
          "execution_count": 9
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}