Challenge with the C3_W1_Assignment

Course Q&A AI For Medicine AI For Medical Treatment
1

Hello everyone,

I’m currently facing a challenge with the C3_W1_Assignment.

My main issue revolves around this particular line of code, which is causing me some trouble:

treatment_model, best_hyperparam_treat = holdout_grid_search(RandomForestClassifier, X_treat_train, y_treat_train, X_treat_val, y_treat_val, hyperparams)

The problem is that the X_treat_train dataset no longer contains the TRTMT column, as it was removed by the treatment_dataset_split function.

However, the c_statistic function (the estimator_score), which is a part of my holdout_grid_search function, requires the TRTMT column to identify the best model.

Here are the parameters needed for the function c_statistic:

c_statistic(pred_rr, y, w, random_seed=0)

where w is an array of true treatments.

The conundrum deepens because, even if X_treat_train included the TRTMT column, filled entirely with ones, it wouldn’t be feasible to split it into two arrays within the c_statistic function - one with TRTMT==1 and another with TRTMT==0 - since we only have TRTMT==1 data.
Consequently, the c_for_benefit_score function, which is a component of c_statistic , fails to function as intended.

This has left me quite perplexed, and I would greatly appreciate any guidance or assistance on this matter.

Best regards,
Alex

2

Hi @Learning_Guy

Welcome! Good to see you on the community.

Not sure if you have already got past this problem, that I am frankly not quite able to understand completely.

X_(anything) is typically the feature set and y_(anything) is typically the label column.

So here…
X_treat_train = X_train[X_train[“TRTMT”]==1]
X_treat_train = X_treat_train.drop(‘TRTMT’,axis=1)
is indeed the correct statement as it removes TRTMT.

And the label column is:
y_treat_train = y_train[X_train[“TRTMT”]==1]

Best Regards.
Jaidev

3

Hello Jaidev,

Thank you very much for your message!

I’d like to describe the still existing problem in more detail for your assistance.

My task involves completing the function c_statistic(pred_rr, y, w, random_seed=0), where w represents an array of true treatments. This function is integral to my assignment and needs to be incorporated into another function, holdout_grid_search.

The holdout_grid_search function requires the following parameters:

  • RandomForestClassifier
  • X_treat_train
  • y_treat_train
  • X_treat_val
  • y_treat_val
  • hyperparams

The challenge arises because X_treat_train and X_treat_val no longer include the w parameter. This leads to my question: How can I effectively integrate c_statistic(pred_rr, y, w, random_seed=0) with X_treat_train and X_treat_val in the holdout_grid_search function?

I greatly appreciate your time and any guidance you can provide on this matter.

Best regards,
Alex

4

Hi, @Learning_Guy

Let me confirm the intent of your question.
Your question is how to treat the argument w in the function c_statistic when w is NOT included in the arguments of the function holdout_grid_search?
Then, it would be difficult to call c_statistic in holdout_grid_search.

If so, can you tell us the relevant part of the code?
We would like to see where c_statistic is called in the holdout_grid_search function.

Kind regards,
Nakamura

5

Hello Nakamura,

Thank you for your message. I appreciate your prompt response. Would it be convenient for you if I send the code via a private message?

Best regards,
Alex

6

Hi, @Learning_Guy

Sorry for the late response.
Sure.

Best regards,
Nakamura