Silvia l Random Forest

[silviagonzalez98]

No description provided.

[ArabellaCM]

Hi Silvia! I've had a look at your random forest/smote upsampling lab.

It looks like you can use the separate parts of code but putting them together hasn't quite worked out.

Remember SMOTE is a process of upsampling your data when you have an unbalanced data set e.g. in this lab you have:

0    90569
1     4843
Name: TARGET_B, dtype: int64

You have upsampled your training sets but you fitted your model to the imbalanced X_train and y_train (instead of X_train_sm and y_train_sm)

When you have an imbalanced data set always remember to be very wary of high scores. It can be useful to look at a confusion matrix or ROC-AUC curve to see how good your results actually are.

Let me know if you have any more questions about upsampling or checking the results of a classification model. They're both important topics :)