feat: sensibility and specificty metrics

[rcsm89]

No description provided.

[cromano8]

Saw you used the confusion matrix from sklearn we have that in SnowflakeML. I saw you wanted to add Sensitivity and Specificity, which are not in SnowflakeML (F1 comes natively). Could you try to create sensitivity and specificity using Snowpark because it'll be a good way for other people to learn? The benefits of this are nothing gets pulled into memory vs sklearn. One way is something like this...

TP = result.filter((F.col("ACT") == 1) & (F.col("PRED") == 1)).count()
FP = result.filter((F.col("ACT") == 0) & (F.col("PRED") == 1)).count()
TN = result.filter((F.col("ACT") == 0) & (F.col("PRED") == 0)).count()
FN = result.filter((F.col("ACT") == 1) & (F.col("PRED") == 0)).count()

then create the formula for the two metrics you want.

[rcsm89]

Hi @cromano8,

Thank you for your suggestion! I've taken your advice and implemented Sensitivity and Specificity metrics using Snowpark within the SnowflakeML environment.

[rcsm89]

@cromano8
what do you think about this version?

[cromano8]

Got our live session using this repo this afternoon, I'll take a closer look later today/tomorrow morning, thanks for the addition

[pythonlearning17]

I don't see the titanic.csv in the folder anymore. + how set up the intro-snowML environment or what all should be present in that for the imports to run ?