How can I use sklearn.inverse transform on my predictions

vision
1

I am currently training a model in pytorch for regression purposes(the model consists of some dense layers).

I have used sklearn’s standard scaler on the data which consists of an image and the ground truth regression targets.
How can I use the scaler’s inverse transform on the predicted output from the model during evaluation stage.
The training code is shown below

for epoch in range(num_epochs):
  model.train()
  for batch_idx,(img,gt_pose) in enumerate(train_loader):
    img=img.to(device)
    gt_pose=gt_pose.to(device)
    pred_pose=model(img)
    train_loss=loss(pred_pose,gt_pose)
    train_loss=train_loss/accum_iter
    train_loss.backward()
    if((batch_idx+1)%accum_iter==0) or (batch_idx+1==len(train_loader)):
      optimizer.step()
      optimizer.zero_grad()
    
    print ('Epoch: %03d/%03d | Batch %04d/%04d | Training loss: %.4f' 
                   %(epoch, num_epochs, batch_idx, 
                     len(train_loader), train_loss))
    train_losses.append(train_loss.item())
  model.eval()
  for batch_idx,(img_val,gt_pose_val) in enumerate(val_loader):
    img_val=img_val.to(device)
    gt_pose_val=gt_pose.to(device)
    pred_pose_val=model(img_val)
    loss_val=loss(pred_pose_val,gt_pose_val)
    print ('Epoch: %03d/%03d | Batch %04d/%04d | Validation loss: %.4f' 
                   %(epoch, num_epochs, batch_idx, 
                     len(train_loader), loss_val))
    valid_losses.append(loss_val.item())

Model evaluation

sc=StandardScaler()
img=img.to(device)#same transformations applied on image as during training
torch.manual_seed(123)
with torch.no_grad():
  pred=model(img)
  pred=pred.detach().cpu().numpy()
#sc.inverse_transform(pred) gives an error

if i do sc.inverse_transform(pred),i get an error of the form scalar needs to be fit on the data.

2

You would need to fit the StandardScaler before using transform or inverse_transform, which is missing in the posted code snippet. Either reuse the original scaler object or store the sc.mean_ and sc.scale_ and undo the normalization manually.

3

I went with storing the mean and scale,and then used them to undo the normalization,thanks