Modifying the thresholds
In the video, you observed how NannyML calculates threshold values and learned how to customize them to suit your solution.
In this exercise, your task is to define two custom standard deviation and custom thresholds and then apply them to the results obtained from the CBPE algorithm for the US Census dataset.
The reference and analysis sets have been pre-loaded as reference
and analysis
, along with the nannyml
library.
This exercise is part of the course
Monitoring Machine Learning in Python
Exercise instructions
- Import
ConstantThreshold
, andStandardDeviationThreshold
fromnannyml.thresholds
. - Initialize the standard deviation method and set
std_lower_multiplier
andstd_upper_multiplier
parameters to2
. - Initialize the constant threshold method and set the lower parameter to
0.9
and upper to0.98
. - Pass the constant threshold method for the
f1
metric and the standard deviation method foraccuracy
to the CBPE algorithm.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Import custom thresholds
from ____.____ import ____, ____
# Initialize custom thresholds
stdt = ____(____=____, ____=____)
ct = ____(____=____, ____=____)
# Initialize the CBPE algorithm
estimator = nannyml.CBPE(
problem_type='classification_binary',
y_pred_proba='predicted_probability',
y_pred='prediction',
y_true='employed',
metrics=['roc_auc', 'accuracy', 'f1'],
thresholds={____: ____, ____: ____})