lstm copy.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "#读取工作簿和工作簿中的工作表\n",
    "data_frame=pd.read_excel('Problem_C_Data_Wordle.xlsx')\n",
    "\n",
    "data = list(reversed(data_frame.loc[ 1: ,\"Unnamed: 4\"].tolist()))\n",
    "\n",
    "start=0\n",
    "datas=data[start:]\n",
    "# for t in range(len(datas)-1):\n",
    "#     datas[t]=datas[t+1]-datas[t]\n",
    "# datas[-1]=data[-1]-data[-2]\n",
    "length=len(datas)\n",
    "days=[[i] for i in range(length)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[90982.93   179.  ] [89094.67      103.63397]\n",
      "Epoch:    0, Loss: 0.50695\n",
      "Epoch:  100, Loss: 0.00321\n",
      "Epoch:  200, Loss: 0.00177\n",
      "Epoch:  300, Loss: 0.00267\n",
      "Epoch:  400, Loss: 0.00136\n",
      "Epoch:  500, Loss: 0.00188\n",
      "Epoch:  600, Loss: 0.00061\n",
      "Epoch:  700, Loss: 0.00105\n",
      "Epoch:  800, Loss: 0.00067\n",
      "Epoch:  900, Loss: 0.00115\n"
     ]
    }
   ],
   "source": [
    "#!/usr/bin/env python\n",
    "# coding: utf-8\n",
    " \n",
    "# In[1]:\n",
    " \n",
    " \n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import torch\n",
    "\n",
    " \n",
    "from torch import nn\n",
    "from torch.autograd import Variable\n",
    " \n",
    " \n",
    "# In[121]:\n",
    " \n",
    " \n",
    "batch_size=32\n",
    "time_step=25\n",
    "epoch=1000\n",
    "input_size=2\n",
    "output_size=1\n",
    "mid_dim=8\n",
    "mid_layers=1\n",
    " \n",
    " \n",
    "# In[122]:\n",
    " \n",
    "class RegLSTM(nn.Module):\n",
    "    def __init__(self, inp_dim, out_dim, mid_dim, mid_layers,batch):\n",
    "        super(RegLSTM, self).__init__()\n",
    " \n",
    "        self.rnn = nn.LSTM(inp_dim, mid_dim, mid_layers,batch_first=batch)  # rnn\n",
    "        self.reg = nn.Sequential(\n",
    "            nn.Linear(mid_dim, mid_dim),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(mid_dim, out_dim),\n",
    "        )  # regression\n",
    " \n",
    "    def forward(self, x):\n",
    "        y = self.rnn(x)[0]  # y, (h, c) = self.rnn(x)\n",
    " \n",
    "        batch_size, seq_len, hid_dim = y.shape\n",
    "        y = y.reshape(-1, hid_dim)\n",
    "        y = self.reg(y)\n",
    "        y = y.reshape(batch_size, seq_len, -1)\n",
    "        return y\n",
    " \n",
    " \n",
    "# In[123]:\n",
    " \n",
    " \n",
    "def load_data():\n",
    "    seq_number=np.array(datas,dtype=np.float32\n",
    "    )\n",
    "    #print(seq_number)\n",
    "    seq_number=seq_number[:,np.newaxis]\n",
    "    \n",
    "    #print(seq_number.shape)\n",
    "    seq_year_month=np.array(days,dtype=np.float32)\n",
    "    #print(seq_year_month.shape)\n",
    "    seq=np.concatenate((seq_number,seq_year_month),axis=1)\n",
    "    print(seq.mean(axis=0),seq.std(axis=0))\n",
    "    seq=(seq-seq.mean(axis=0))/seq.std(axis=0)\n",
    "    return(seq)\n",
    " \n",
    " \n",
    "# In[124]:\n",
    " \n",
    " \n",
    "data=load_data()\n",
    "#print(data)\n",
    "train_size=int(len(data)-1)\n",
    " \n",
    " \n",
    "# In[125]:\n",
    " \n",
    " \n",
    "data_sample=np.zeros((train_size-time_step+1,time_step,input_size))\n",
    "label_sample=np.zeros((train_size-time_step+1,time_step,output_size))\n",
    "for i in range(train_size-time_step+1):\n",
    "    data_sample[i]=data[i:i+time_step,:]\n",
    "    label_sample[i]=data[i+1:i+1+time_step,0:1:]\n",
    " \n",
    " \n",
    "# In[126]:\n",
    " \n",
    " \n",
    "device=torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "net=RegLSTM(input_size,output_size,mid_dim,mid_layers,True).to(device)\n",
    "criterion=nn.MSELoss()\n",
    "optimizer=torch.optim.Adam(net.parameters(),lr=1e-2)\n",
    " \n",
    " \n",
    "# In[127]:\n",
    " \n",
    " \n",
    "for i in range(epoch):\n",
    "    for j in range(int((train_size-time_step+1)/batch_size)):\n",
    "        train_X=data_sample[j*batch_size:(j+1)*batch_size,:,:]\n",
    "        train_Y=label_sample[j*batch_size:(j+1)*batch_size,:,:]\n",
    "        var_x=torch.tensor(train_X,dtype=torch.float32,device=device)\n",
    "        var_y=torch.tensor(train_Y,dtype=torch.float32,device=device)\n",
    "        out = net(var_x)\n",
    "        loss=criterion(out,var_y)\n",
    "        #loss = criterion(out[:,-1,:], var_y[:,-1,:])\n",
    "        optimizer.zero_grad()\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "    train_X=data_sample[(j+1)*batch_size:,:,:]\n",
    "    train_Y=label_sample[(j+1)*batch_size:,:,:]\n",
    "    var_x=torch.tensor(train_X,dtype=torch.float32,device=device)\n",
    "    var_y=torch.tensor(train_Y,dtype=torch.float32,device=device)\n",
    "    out = net(var_x)\n",
    "    loss = criterion(out, var_y)\n",
    "    optimizer.zero_grad()\n",
    "    loss.backward()\n",
    "    optimizer.step()\n",
    "    if i%100==0:\n",
    "        print('Epoch: {:4}, Loss: {:.5f}'.format(i, loss.item()))\n",
    " \n",
    " \n",
    "# In[128]:\n",
    " \n",
    " \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def func(x, a, b, c,d):\n",
    "    return a * np.exp(-(x-d)*b) + c\n",
    "popt_min=np.array([ 4.34703147e+05 , 1.78907252e-02 , 2.28909992e+04 ,-1.99427168e+01])\n",
    "popt_max=np.array([ 4.49901532e+05,  1.76663163e-02 , 2.43997347e+04 ,-1.48954702e+01])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(359, 2)\n",
      "[20719.844]\n",
      "[23037.83330171]\n",
      "[20250.578]\n",
      "[22868.57716581]\n",
      "[20186.406]\n",
      "[22834.80033831]\n",
      "[19996.664]\n",
      "[22759.59915555]\n",
      "[19862.781]\n",
      "[22703.43508337]\n",
      "[19704.992]\n",
      "[22639.70501051]\n",
      "[19557.96]\n",
      "[22579.94956357]\n",
      "[19406.992]\n",
      "[22519.25673506]\n",
      "[19258.54]\n",
      "[22459.76447044]\n",
      "[19110.14]\n",
      "[22400.63981821]\n",
      "[18962.805]\n",
      "[22342.20651819]\n",
      "[18816.11]\n",
      "[22284.31217353]\n",
      "[18670.352]\n",
      "[22227.04438171]\n",
      "[18525.328]\n",
      "[22170.3244702]\n",
      "[18381.227]\n",
      "[22114.20435758]\n",
      "[18237.883]\n",
      "[22058.6191444]\n",
      "[18095.398]\n",
      "[22003.59282937]\n",
      "[17953.68]\n",
      "[21949.08465113]\n",
      "[17812.703]\n",
      "[21895.0776179]\n",
      "[17672.617]\n",
      "[21841.61235004]\n",
      "[17533.352]\n",
      "[21788.65686052]\n",
      "[17394.836]\n",
      "[21736.17946081]\n",
      "[17257.04]\n",
      "[21684.16177128]\n",
      "[17120.07]\n",
      "[21632.63256531]\n",
      "[16983.938]\n",
      "[21581.58703041]\n",
      "[16848.484]\n",
      "[21530.96592554]\n",
      "[16713.867]\n",
      "[21480.8144262]\n",
      "[16579.93]\n",
      "[21431.07378242]\n",
      "[16446.71]\n",
      "[21381.75058113]\n",
      "[16314.383]\n",
      "[21332.89590471]\n",
      "[16182.719]\n",
      "[21284.42823984]\n",
      "[16051.75]\n",
      "[21236.35221991]\n",
      "[15921.555]\n",
      "[21188.68830521]\n",
      "[15792.117]\n",
      "[21141.42590089]\n",
      "[15663.359]\n",
      "[21094.53376683]\n",
      "[15535.359]\n",
      "[21048.03292772]\n",
      "[15408.016]\n",
      "[21001.88468759]\n",
      "[15281.508]\n",
      "[20956.14426949]\n",
      "[15155.516]\n",
      "[20910.70039295]\n",
      "[15030.422]\n",
      "[20865.67630994]\n",
      "[14905.883]\n",
      "[20820.95323686]\n",
      "[14782.117]\n",
      "[20776.60003677]\n",
      "[14658.953]\n",
      "[20732.55550592]\n",
      "[14536.531]\n",
      "[20688.86274377]\n",
      "[14414.789]\n",
      "[20645.49727329]\n",
      "[14293.648]\n",
      "[20602.42953632]\n",
      "[14173.25]\n",
      "[20559.70301413]\n",
      "[14053.516]\n",
      "[20517.28838979]\n",
      "[13934.352]\n",
      "[20475.15125241]\n",
      "[13815.914]\n",
      "[20433.34063476]\n",
      "[13698.141]\n",
      "[20391.83275945]\n",
      "[13581.039]\n",
      "[20350.62738938]\n",
      "[13464.477]\n",
      "[20309.67751164]\n",
      "[13348.633]\n",
      "[20269.04037571]\n",
      "[13233.305]\n",
      "[20228.6457193]\n",
      "[13118.789]\n",
      "[20188.59003593]\n",
      "[13004.75]\n",
      "[20148.75896769]\n",
      "[12891.406]\n",
      "[20109.22313555]\n",
      "[12778.594]\n",
      "[20069.92563644]\n",
      "[12666.547]\n",
      "[20030.94245715]\n",
      "[[-0.76095191]\n",
      " [-0.7611793 ]\n",
      " [-0.75990026]\n",
      " [-0.75909009]\n",
      " [-0.75807503]\n",
      " [-0.75715835]\n",
      " [-0.75620508]\n",
      " [-0.75526016]\n",
      " [-0.75432353]\n",
      " [-0.75338385]\n",
      " [-0.75244108]\n",
      " [-0.75150643]\n",
      " [-0.75056855]\n",
      " [-0.74963867]\n",
      " [-0.74870544]\n",
      " [-0.7477801 ]\n",
      " [-0.74685127]\n",
      " [-0.74591893]\n",
      " [-0.74499431]\n",
      " [-0.74407737]\n",
      " [-0.7431567 ]\n",
      " [-0.74223226]\n",
      " [-0.74131535]\n",
      " [-0.74040589]\n",
      " [-0.73949247]\n",
      " [-0.73858641]\n",
      " [-0.73767627]\n",
      " [-0.73676203]\n",
      " [-0.73586637]\n",
      " [-0.73496646]\n",
      " [-0.73406226]\n",
      " [-0.73316513]\n",
      " [-0.73227499]\n",
      " [-0.73138038]\n",
      " [-0.73049269]\n",
      " [-0.72960042]\n",
      " [-0.7287264 ]\n",
      " [-0.72783624]\n",
      " [-0.72696425]\n",
      " [-0.72608746]\n",
      " [-0.72521729]\n",
      " [-0.72434222]\n",
      " [-0.72347368]\n",
      " [-0.72261163]\n",
      " [-0.72174451]\n",
      " [-0.72088379]\n",
      " [-0.72002942]\n",
      " [-0.71916984]\n",
      " [-0.71831652]\n",
      " [-0.71746942]\n",
      " [-0.71662849]\n",
      " [-0.71578214]\n",
      " [-0.71494187]\n",
      " [-0.71409611]\n",
      " [-0.71326791]\n",
      " [-0.7124341 ]\n",
      " [-0.7116062 ]\n",
      " [-0.71077261]\n",
      " [-0.70995645]\n",
      " [-0.70913448]]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f24ac35a550>]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "net=net.eval()\n",
    "\n",
    "test_y=list()\n",
    "reals=list()\n",
    "new_mean=np.array([90982.93  , 179.   ])\n",
    "new_std=np.array([89094.67    ,  103.63397])\n",
    "print(data.shape) \n",
    "for i in range(60):\n",
    "    test_X=data[-time_step:,:]\n",
    "    test_x=test_X[:time_step,:].reshape(1,time_step,input_size)\n",
    "    test_x=torch.tensor(test_x,dtype=torch.float32,device=device)\n",
    "    pre_data=len(data)+i\n",
    "    y_predict_max=[func(pre_data, popt_max[0],popt_max[1],popt_max[2],popt_max[3])]\n",
    "    y_predict_min=[func(pre_data, popt_min[0],popt_min[1],popt_min[2],popt_min[3])]\n",
    "    tem=net(test_x).cpu().data.numpy()\n",
    "    test_y.append(tem[0][-1])\n",
    "    print(test_y[-1]*new_std[0]+new_mean[0])\n",
    "    for item in data:\n",
    "        item[0]=item[0]*new_std[0]+new_mean[0]\n",
    "        item[1]=item[1]*new_std[1]+new_mean[1]\n",
    "    \n",
    "    real=test_y[-1]*new_std[0]+new_mean[0]\n",
    "    real=(real+y_predict_min+y_predict_max)/3\n",
    "    print(real)\n",
    "    \n",
    "    new=np.array([[int(real),train_size+1+i]])\n",
    "    data=np.concatenate((data,new),axis=0)\n",
    "    \n",
    "    new_mean=data.mean(axis=0)\n",
    "    new_std=data.std(axis=0)\n",
    "    data=(data-data.mean(axis=0))/data.std(axis=0)\n",
    "    reals.append(data[-1][0])\n",
    "    #print(data)\n",
    "test_y=np.array(reals).reshape((-1,1))\n",
    "\n",
    "print(test_y)\n",
    "\n",
    "\n",
    "\n",
    "plt.plot(test_y, 'r', label='pred')\n",
    "\n",
    " \n",
    " \n",
    "\n",
    " \n",
    " \n",
    " \n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[85998.89268236   120.954536  ] [81014.78008931   209.00000068]\n"
     ]
    }
   ],
   "source": [
    "print(new_std,new_mean)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[20030.]\n"
     ]
    }
   ],
   "source": [
    "print(test_y[-1]*new_std[0]+new_mean[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "bdec6ca5ce638a4776cd655f71204534c127e30145c88be198b9bebdb38fc0db"
  },
  "kernelspec": {
   "display_name": "Python 3.6.5 ('snow')",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}