Exam DP-100 topic 2 question 49 discussion

Stratified split should be true in order to be balanced

In my opinion, stratified split should be true. Based on this: "Stratified split: Set this option to True to ensure that the two output datasets contain a representative sample of the values in the strata column or stratification key column. With stratified sampling, the data is divided such that each output dataset gets roughly the same percentage of each target value. For example, you might want to ensure that your training and testing sets are roughly balanced with regard to the outcome, or with regard to some other column such as gender." https://docs.microsoft.com/en-us/azure/machine-learning/studio-module-reference/split-data-using-split-rows

When performing data splitting, especially in scenarios where there is an imbalance in the distribution of certain categories or groups within a column, using a strata column can be beneficial. By grouping the rows based on the strata column, it helps ensure that each subset (e.g., training or testing) maintains a similar representation of the different categories or groups present in the strata column. so Stratified split - TRUE SaM

Stratified split should be FALSE, because it is not the IMBALANCE DATASET. Please refer this: Chawla, Nitesh V., et al. "SMOTE: synthetic minority over-sampling technique." Journal of Artificial Intelligence Research 16 (2002): 321-357.

In a classification setting, it is often chosen to ensure that the train and test sets have approximately the same percentage of samples of each target class as the complete set. As a result, if the data set has a large amount of each class, stratified sampling is pretty much the same as random sampling. But if one class isn't much represented in the data set, which may be the case in your dataset since you plan to oversample the minority class, then stratified sampling may yield a different target class distribution in the train and test sets than what random sampling may yield.

In a classification setting, it is often chosen to ensure that the train and test sets have approximately the same percentage of samples of each target class as the complete set. As a result, if the data set has a large amount of each class, stratified sampling is pretty much the same as random sampling. But if one class isn't much represented in the data set, which may be the case in your dataset since you plan to oversample the minority class, then stratified sampling may yield a different target class distribution in the train and test sets than what random sampling may yield.

I think Stratified split should be set to True for balancing the 2 subsets, no info about Random Splitting True/False though...

Based on below requirement I would say Stratified Split should be false Additional requirements for stratified sampling: The strata column can contain only nominal or categorical data. If the column contains continuous numeric data, an error message is raised. A column with too many unique values is not a good candidate for stratification. You might try collapsing some categories or grouping values beforehand.

Contrary to all comments here, stratified split should be False. For it to be set to true, we need to select the column for which data needs to be stratified. There is no mention of any column in the question. In this scenario just saying select it to True does not make sense. Also for people following documentation, here is the extract. Stratified split: Set this option to True to ensure that the two output datasets contain a representative sample of the values in the strata column or stratification key column. With stratified sampling, the data is divided such that each output dataset gets roughly the same percentage of each target value. For example, you might want to ensure that your training and testing sets are roughly balanced with regard to the outcome or to some other column (such as gender).

Since, the dataset is balanced, Stratified split = False might work.

I too think Stratified should be set to True.

stratified split should be set to True With stratified sampling, the data is divided such that each output dataset gets roughly the same percentage of each target value. For example, you might want to ensure that your training and testing sets are roughly balanced with regard to the outcome, or with regard to some other column such as gender. https://docs.microsoft.com/en-us/azure/machine-learning/studio-module-reference/split-data-using-split-rows

Stratified should be true

Stratified means UNbalanced as it keeps the ratio between classes. That is why it needs to be set to False.

because of balanced testing and training samples, Statisfied should be set to True. https://docs.microsoft.com/en-us/azure/machine-learning/studio-module-reference/split-data-using-split-rows

Stratified split is correct here

I too think Statified should be set to True