Pandas DF 專換訓練ML 模型 使用 Pytorch 完整流程範例

 從 Pandas 資料 ➜ 標準化 ➜ 分群 ➜ 建立 PyTorch Dataset ➜ 模型訓練。

我會以下列順序一步步教你完整流程範例。

---

✅ 假設你已經有：

python
type0data, type1data, type2data  # NumPy 陣列，每個 shape=(N, 6)，特徵為 b0~b5

---

🧱 Step 1：組合所有類別 & 建立 label

python
import numpy as np

# 將所有資料組合起來
X = np.vstack([type0data, type1data, type2data])

# 建立 label（0, 1, 2）
y = np.concatenate([
    np.zeros(len(type0data)),
    np.ones(len(type1data)),
    np.full(len(type2data), 2)
])

print(X.shape)  # (total_samples, 6)
print(y.shape)  # (total_samples,)

---

🧼 Step 2：資料標準化（Zero-mean, Unit-std）

python
from sklearn.preprocessing import StandardScaler

scaler = StandardScaler()
X_scaled = scaler.fit_transform(X)  # 注意：要先 fit，再 transform

---

✂️ Step 3：分群 (Train/Test Split)

python
from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(X_scaled, y, test_size=0.2, random_state=42, stratify=y)

---

🔄 Step 4：轉成 PyTorch Dataset

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

# 轉成 tensor
X_train_tensor = torch.tensor(X_train, dtype=torch.float32)
y_train_tensor = torch.tensor(y_train, dtype=torch.long)

X_test_tensor = torch.tensor(X_test, dtype=torch.float32)
y_test_tensor = torch.tensor(y_test, dtype=torch.long)

# 建立 PyTorch Dataset
train_dataset = TensorDataset(X_train_tensor, y_train_tensor)
test_dataset = TensorDataset(X_test_tensor, y_test_tensor)

# 建立 DataLoader（for batch training）
train_loader = DataLoader(train_dataset, batch_size=32, shuffle=True)
test_loader = DataLoader(test_dataset, batch_size=32)

---

🧠 Step 5：簡單模型訓練範例

python
import torch.nn as nn
import torch.optim as optim

# 建立簡單 MLP 模型
class SimpleNet(nn.Module):
    def __init__(self):
        super(SimpleNet, self).__init__()
        self.model = nn.Sequential(
            nn.Linear(6, 32),
            nn.ReLU(),
            nn.Linear(32, 3)  # 3 類別
        )

    def forward(self, x):
        return self.model(x)

model = SimpleNet()
criterion = nn.CrossEntropyLoss()
optimizer = optim.Adam(model.parameters(), lr=0.001)

---

🚀 Step 6：訓練模型

python
for epoch in range(20):
    model.train()
    total_loss = 0
    for xb, yb in train_loader:
        pred = model(xb)
        loss = criterion(pred, yb)

        optimizer.zero_grad()
        loss.backward()
        optimizer.step()

        total_loss += loss.item()
    
    print(f"Epoch {epoch+1}, Loss: {total_loss:.4f}")

---

🧪 Step 7：測試模型

python
model.eval()
correct = 0
total = 0
with torch.no_grad():
    for xb, yb in test_loader:
        pred = model(xb)
        predicted = torch.argmax(pred, dim=1)
        correct += (predicted == yb).sum().item()
        total += yb.size(0)

print(f"Accuracy: {correct / total:.2%}")

---

✅ 小結：

步驟	說明
標準化	使用 StandardScaler 處理各特徵
分群	使用 train_test_split 並保留類別分布
TensorDataset	將 X, y 包裝成可被 PyTorch 處理的資料集
模型	用 nn.Sequential 建構簡單分類器
訓練	搭配 CrossEntropyLoss + Adam 優化器