AttributeError: 'float' object has no attribute 'item'

[ayyucedemirbas]

Hello,

After running this cell in the Tutorial 1:

benchmark_expression_conversion(
    benchmark_output_dir=output_path,
    save_name=transformation_benchmarking_save_name,
    normalized_expression_matrix=adata.X,
    sample_size=1024,
)

I encounter this error message:

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
[<ipython-input-29-808987c0b700>](https://localhost:8080/#) in <cell line: 1>()
----> 1 benchmark_expression_conversion(
      2     benchmark_output_dir=output_path,
      3     save_name=transformation_benchmarking_save_name,
      4     normalized_expression_matrix=adata.X,
      5     sample_size=1024,

[/usr/local/lib/python3.10/dist-packages/cell2sentence/utils.py](https://localhost:8080/#) in benchmark_expression_conversion(benchmark_output_dir, save_name, normalized_expression_matrix, sample_size)
    278         "slope": [linear_reg.coef_.item()],
    279         "intercept": [linear_reg.intercept_.item()],
--> 280         "r_squared": [r_squared_score.item()],
    281         "pearson_r_statistic": [pearson_r_score.statistic.item()],
    282         "pearson_r_pvalue": [pearson_r_score.pvalue.item()],

AttributeError: 'float' object has no attribute 'item'

It seems the error is caused by trying to call .item() on a float. Instead, the direct float values could be used, or checks could be added to ensure the correct data type before calling .item().

Thank you so much!

[ayyucedemirbas]

I have submitted a pull request to solve this problem. Thanks.

[SyedA5688]

Hi there,

Thank you for pointing this out and for making a pull request about the issue, we greatly appreciate the open-source contribution!

We've noticed similar AttributeErrors before depending on the version of certain packages used to compute these correlation scores (scipy, sklearn are the usual culprits). I will do some testing of package versions on my end, and afterwards will take another look at your pull request. Could you let me know what version of scipy and sklearn you have in your environment? Thanks again!

[ayyucedemirbas]

Hi,

Thank you so much for your time and attention.

I'm using Google Colab, which has scipy 1.13.1 and sklearn 1.5.2.

Thanks.