
import numpy as np
import torch
import torch.nn as nn
import torch.nn.functional as F


from torch.utils.data import TensorDataset
from torch.utils.data import DataLoader


X_train  =  torch.FloatTensor([[73,  80,  75], 
                               [93,  88,  93], 
                               [89,  91,  90], 
                               [96,  98,  100],   
                               [73,  66,  70]])  
y_train  =  torch.FloatTensor([[152],  [185],  [180],  [196],  [142]])


# TensorDataset은 기본적으로 텐서를 입력
dataset = TensorDataset(X_train, y_train)


# 데이터로더는 기본적으로 2개의 인자를 입력받는다. 
# 하나는 데이터셋, 미니 배치의 크기. 미니 배치의 크기는 2의 배수 사용
# shuffle=True를 선택하면 Epoch마다 데이터셋을 섞어 학습 순서를 바꿈
dataloader = DataLoader(dataset, batch_size=2, shuffle=True)


# 모델과 옵티마이저 설계
model = nn.Linear(3,1)
optimizer = torch.optim.SGD(model.parameters(), lr=1e-5)


nb_epochs = 40
for epoch in range(nb_epochs + 1):
  for batch_idx, samples in enumerate(dataloader):
    # print(batch_idx)
    # print(samples)
    x_train, y_train = samples
    # H(x) 계산
    prediction = model(x_train)

    # cost 계산
    cost = F.mse_loss(prediction, y_train)

    # cost로 H(x) 계산
    optimizer.zero_grad()
    cost.backward()
    optimizer.step()

    print('Epoch {:4d}/{} Batch {}/{} Cost: {:.6f}'.format(
        epoch, nb_epochs, batch_idx+1, len(dataloader),
        cost.item()
        ))

Epoch    0/40 Batch 1/3 Cost: 109422.382812
Epoch    0/40 Batch 2/3 Cost: 13926.935547
Epoch    0/40 Batch 3/3 Cost: 9553.985352
Epoch    1/40 Batch 1/3 Cost: 2407.589355
Epoch    1/40 Batch 2/3 Cost: 630.397583
Epoch    1/40 Batch 3/3 Cost: 332.481873
Epoch    2/40 Batch 1/3 Cost: 54.563324
Epoch    2/40 Batch 2/3 Cost: 15.682707
Epoch    2/40 Batch 3/3 Cost: 15.891177
Epoch    3/40 Batch 1/3 Cost: 2.497978
Epoch    3/40 Batch 2/3 Cost: 1.184473
Epoch    3/40 Batch 3/3 Cost: 0.711837
Epoch    4/40 Batch 1/3 Cost: 0.044865
Epoch    4/40 Batch 2/3 Cost: 1.406529
Epoch    4/40 Batch 3/3 Cost: 2.957722
Epoch    5/40 Batch 1/3 Cost: 1.657993
Epoch    5/40 Batch 2/3 Cost: 1.495775
Epoch    5/40 Batch 3/3 Cost: 1.046519
Epoch    6/40 Batch 1/3 Cost: 1.526840
Epoch    6/40 Batch 2/3 Cost: 1.215872
Epoch    6/40 Batch 3/3 Cost: 1.332472
Epoch    7/40 Batch 1/3 Cost: 1.218087
Epoch    7/40 Batch 2/3 Cost: 1.446331
Epoch    7/40 Batch 3/3 Cost: 0.300509
Epoch    8/40 Batch 1/3 Cost: 0.889232
Epoch    8/40 Batch 2/3 Cost: 1.583050
Epoch    8/40 Batch 3/3 Cost: 0.173692
Epoch    9/40 Batch 1/3 Cost: 0.295575
Epoch    9/40 Batch 2/3 Cost: 1.129346
Epoch    9/40 Batch 3/3 Cost: 2.637502
Epoch   10/40 Batch 1/3 Cost: 0.944092
Epoch   10/40 Batch 2/3 Cost: 1.281095
Epoch   10/40 Batch 3/3 Cost: 1.154995
Epoch   11/40 Batch 1/3 Cost: 0.160964
Epoch   11/40 Batch 2/3 Cost: 1.212293
Epoch   11/40 Batch 3/3 Cost: 2.744630
Epoch   12/40 Batch 1/3 Cost: 1.786660
Epoch   12/40 Batch 2/3 Cost: 1.454112
Epoch   12/40 Batch 3/3 Cost: 1.002473
Epoch   13/40 Batch 1/3 Cost: 1.289761
Epoch   13/40 Batch 2/3 Cost: 1.020319
Epoch   13/40 Batch 3/3 Cost: 1.456298
Epoch   14/40 Batch 1/3 Cost: 0.453115
Epoch   14/40 Batch 2/3 Cost: 3.129420
Epoch   14/40 Batch 3/3 Cost: 0.284713
Epoch   15/40 Batch 1/3 Cost: 0.882871
Epoch   15/40 Batch 2/3 Cost: 1.571089
Epoch   15/40 Batch 3/3 Cost: 0.166343
Epoch   16/40 Batch 1/3 Cost: 0.915380
Epoch   16/40 Batch 2/3 Cost: 0.512459
Epoch   16/40 Batch 3/3 Cost: 3.065117
Epoch   17/40 Batch 1/3 Cost: 1.628652
Epoch   17/40 Batch 2/3 Cost: 0.476308
Epoch   17/40 Batch 3/3 Cost: 1.290222
Epoch   18/40 Batch 1/3 Cost: 2.241758
Epoch   18/40 Batch 2/3 Cost: 1.175711
Epoch   18/40 Batch 3/3 Cost: 0.508409
Epoch   19/40 Batch 1/3 Cost: 1.225560
Epoch   19/40 Batch 2/3 Cost: 0.207209
Epoch   19/40 Batch 3/3 Cost: 3.247816
Epoch   20/40 Batch 1/3 Cost: 0.790679
Epoch   20/40 Batch 2/3 Cost: 1.899174
Epoch   20/40 Batch 3/3 Cost: 2.129081
Epoch   21/40 Batch 1/3 Cost: 1.683914
Epoch   21/40 Batch 2/3 Cost: 0.716574
Epoch   21/40 Batch 3/3 Cost: 0.859389
Epoch   22/40 Batch 1/3 Cost: 1.073815
Epoch   22/40 Batch 2/3 Cost: 1.591716
Epoch   22/40 Batch 3/3 Cost: 0.101120
Epoch   23/40 Batch 1/3 Cost: 0.504671
Epoch   23/40 Batch 2/3 Cost: 1.815940
Epoch   23/40 Batch 3/3 Cost: 1.404979
Epoch   24/40 Batch 1/3 Cost: 1.258580
Epoch   24/40 Batch 2/3 Cost: 1.488407
Epoch   24/40 Batch 3/3 Cost: 1.007675
Epoch   25/40 Batch 1/3 Cost: 1.269281
Epoch   25/40 Batch 2/3 Cost: 0.724787
Epoch   25/40 Batch 3/3 Cost: 2.530030
Epoch   26/40 Batch 1/3 Cost: 0.429210
Epoch   26/40 Batch 2/3 Cost: 1.682514
Epoch   26/40 Batch 3/3 Cost: 1.233186
Epoch   27/40 Batch 1/3 Cost: 0.386668
Epoch   27/40 Batch 2/3 Cost: 1.869890
Epoch   27/40 Batch 3/3 Cost: 1.461754
Epoch   28/40 Batch 1/3 Cost: 0.921992
Epoch   28/40 Batch 2/3 Cost: 1.965172
Epoch   28/40 Batch 3/3 Cost: 1.559793
Epoch   29/40 Batch 1/3 Cost: 0.889630
Epoch   29/40 Batch 2/3 Cost: 1.770577
Epoch   29/40 Batch 3/3 Cost: 1.301199
Epoch   30/40 Batch 1/3 Cost: 0.196662
Epoch   30/40 Batch 2/3 Cost: 3.392961
Epoch   30/40 Batch 3/3 Cost: 0.213224
Epoch   31/40 Batch 1/3 Cost: 0.876909
Epoch   31/40 Batch 2/3 Cost: 1.136801
Epoch   31/40 Batch 3/3 Cost: 1.421413
Epoch   32/40 Batch 1/3 Cost: 1.700405
Epoch   32/40 Batch 2/3 Cost: 0.732435
Epoch   32/40 Batch 3/3 Cost: 1.076102
Epoch   33/40 Batch 1/3 Cost: 0.065904
Epoch   33/40 Batch 2/3 Cost: 3.042134
Epoch   33/40 Batch 3/3 Cost: 1.361230
Epoch   34/40 Batch 1/3 Cost: 1.622181
Epoch   34/40 Batch 2/3 Cost: 0.974258
Epoch   34/40 Batch 3/3 Cost: 0.626623
Epoch   35/40 Batch 1/3 Cost: 1.162878
Epoch   35/40 Batch 2/3 Cost: 1.340035
Epoch   35/40 Batch 3/3 Cost: 0.813756
Epoch   36/40 Batch 1/3 Cost: 1.067085
Epoch   36/40 Batch 2/3 Cost: 0.347904
Epoch   36/40 Batch 3/3 Cost: 3.170678
Epoch   37/40 Batch 1/3 Cost: 0.878515
Epoch   37/40 Batch 2/3 Cost: 1.548400
Epoch   37/40 Batch 3/3 Cost: 0.206153
Epoch   38/40 Batch 1/3 Cost: 1.446815
Epoch   38/40 Batch 2/3 Cost: 0.541471
Epoch   38/40 Batch 3/3 Cost: 1.373899
Epoch   39/40 Batch 1/3 Cost: 1.694149
Epoch   39/40 Batch 2/3 Cost: 0.997300
Epoch   39/40 Batch 3/3 Cost: 0.072738
Epoch   40/40 Batch 1/3 Cost: 0.287293
Epoch   40/40 Batch 2/3 Cost: 1.123619
Epoch   40/40 Batch 3/3 Cost: 2.561045


new_var =  torch.FloatTensor([[73, 80, 75]])
pred_y = model(new_var) 
print("훈련 후 입력이 73, 80, 75일 때의 예측값 :", pred_y)

훈련 후 입력이 73, 80, 75일 때의 예측값 : tensor([[152.7064]], grad_fn=<AddmmBackward>)

[NLP, sklearn] 호텔스 닷컴 사용자 후기 및 평점 분석, 댓글 감성 분석 예측 모델 - Naive Bayes Classification (1)	2021.02.19
[NLP, gensim] 영문 데이터 Word2Vec 만들고 Embedding Visualization으로 시각화 하기 (0)	2021.02.17
[PyTorch] 파이토치 기본 조작2 - view, Squeeze, Unsqueeze, Type Casting, concatenate, Stacking, ones_like와 zeros_like, In-place Operation (0)	2021.02.13
[PyTorch] 파이토치 기본 조작 - Tensor 만들기, .matmul, .mul, 브로드캐스팅, mean(), sum(), max(), agrmax() (0)	2021.02.08
[sklearn, KMeans] Kmeans Clustering 을 활용한 데이터 기반 고객 분류 (0)	2021.02.01

Data Navigator

Data Navigator

[PyTorch] Mini Batch and Data Load를 이용한 다중 회귀 분석 모델 본문

[PyTorch] Mini Batch and Data Load를 이용한 다중 회귀 분석 모델

예측값 152.7064 / 정답은 152¶

'Machine Learning, Deep Learning' 카테고리의 다른 글

티스토리툴바

« 2024/12 »
일	월	화	수	목	금	토
1	2	3	4	5	6	7
8	9	10	11	12	13	14
15	16	17	18	19	20	21
22	23	24	25	26	27	28
29	30	31