Increasing learning rate

[rabah-khalek]

Increasing learning rate from 0.1 to 0.2.

[github-actions]

Hi, this is a report generated by the Giskard scan 🐢.

We have identified potential vulnerabilities in your model based on an automated scan.

However, it's important to note that automated scans may produce false positives or miss certain vulnerabilities. We encourage you to review the findings and assess the impact accordingly.

Report summary

Performance issues (12)

Vulnerability	Level	Data slice	Metric	Transformation	Deviation	Description
Performance	major	Hospital_code == 26	Precision = 0.371	—	-12.44% than global	For records in your dataset where Hospital_code == 26, the Precision is 12.44% lower than the global Precision.
Performance	major	City_Code_Hospital == 2	Precision = 0.372	—	-12.37% than global	For records in your dataset where City_Code_Hospital == 2, the Precision is 12.37% lower than the global Precision.
Performance	major	Ward_Facility_Code == "D"	Precision = 0.372	—	-12.37% than global	For records in your dataset where Ward_Facility_Code == "D", the Precision is 12.37% lower than the global Precision.
Performance	major	Hospital_code == 11	Precision = 0.373	—	-12.10% than global	For records in your dataset where Hospital_code == 11, the Precision is 12.1% lower than the global Precision.
Performance	medium	Ward_Facility_Code == "A"	Precision = 0.384	—	-9.38% than global	For records in your dataset where Ward_Facility_Code == "A", the Precision is 9.38% lower than the global Precision.
Performance	medium	City_Code_Hospital == 3	Precision = 0.388	—	-8.56% than global	For records in your dataset where City_Code_Hospital == 3, the Precision is 8.56% lower than the global Precision.
Performance	medium	Hospital_type_code == "b"	Precision = 0.391	—	-7.72% than global	For records in your dataset where Hospital_type_code == "b", the Precision is 7.72% lower than the global Precision.
Performance	medium	Hospital_type_code == "c"	Precision = 0.394	—	-7.07% than global	For records in your dataset where Hospital_type_code == "c", the Precision is 7.07% lower than the global Precision.
Performance	medium	Ward_Type == "S"	Precision = 0.396	—	-6.70% than global	For records in your dataset where Ward_Type == "S", the Precision is 6.7% lower than the global Precision.
Performance	medium	City_Code_Patient == 5.000	Precision = 0.396	—	-6.64% than global	For records in your dataset where City_Code_Patient == 5.000, the Precision is 6.64% lower than the global Precision.
Performance	medium	Type of Admission == "Emergency"	Precision = 0.400	—	-5.60% than global	For records in your dataset where Type of Admission == "Emergency", the Precision is 5.6% lower than the global Precision.
Performance	medium	Hospital_region_code == "Z"	Precision = 0.402	—	-5.28% than global	For records in your dataset where Hospital_region_code == "Z", the Precision is 5.28% lower than the global Precision.

Overconfidence issues (3)

Vulnerability	Level	Data slice	Metric	Transformation	Deviation	Description
Overconfidence	medium	Bed Grade == 2.000	Overconfidence rate = 0.630	—	+17.73% than global	For records in your dataset where Bed Grade == 2.000, we found a significantly higher number of overconfident wrong predictions (8541 samples, corresponding to 63.033210332103316% of the wrong predictions in the data slice).
Overconfidence	medium	City_Code_Hospital == 7	Overconfidence rate = 0.626	—	+16.95% than global	For records in your dataset where City_Code_Hospital == 7, we found a significantly higher number of overconfident wrong predictions (2336 samples, corresponding to 62.61056017153578% of the wrong predictions in the data slice).
Overconfidence	medium	Ward_Facility_Code == "C"	Overconfidence rate = 0.626	—	+16.95% than global	For records in your dataset where Ward_Facility_Code == "C", we found a significantly higher number of overconfident wrong predictions (2336 samples, corresponding to 62.61056017153578% of the wrong predictions in the data slice).

Underconfidence issues (13)

Vulnerability	Level	Data slice	Metric	Transformation	Deviation	Description
Underconfidence	major	City_Code_Patient == 5.000	Overconfidence rate = 0.084	—	+34.11% than global	For records in your dataset where City_Code_Patient == 5.000, we found a significantly higher number of underconfident predictions (337 samples, corresponding to 8.4% of the predictions in the data slice).
Underconfidence	major	Bed Grade == 3.000	Overconfidence rate = 0.084	—	+33.84% than global	For records in your dataset where Bed Grade == 3.000, we found a significantly higher number of underconfident predictions (1864 samples, corresponding to 8.4% of the predictions in the data slice).
Underconfidence	major	Hospital_code == 11	Overconfidence rate = 0.084	—	+32.87% than global	For records in your dataset where Hospital_code == 11, we found a significantly higher number of underconfident predictions (287 samples, corresponding to 8.4% of the predictions in the data slice).
Underconfidence	major	Bed Grade == 4.000	Overconfidence rate = 0.083	—	+31.46% than global	For records in your dataset where Bed Grade == 4.000, we found a significantly higher number of underconfident predictions (958 samples, corresponding to 8.3% of the predictions in the data slice).
Underconfidence	major	Severity of Illness == "Minor"	Overconfidence rate = 0.080	—	+27.09% than global	For records in your dataset where Severity of Illness == "Minor", we found a significantly higher number of underconfident predictions (1366 samples, corresponding to 8.0% of the predictions in the data slice).
Underconfidence	major	City_Code_Hospital == 5	Overconfidence rate = 0.076	—	+20.88% than global	For records in your dataset where City_Code_Hospital == 5, we found a significantly higher number of underconfident predictions (475 samples, corresponding to 7.6% of the predictions in the data slice).
Underconfidence	medium	Hospital_type_code == "e"	Overconfidence rate = 0.075	—	+19.12% than global	For records in your dataset where Hospital_type_code == "e", we found a significantly higher number of underconfident predictions (374 samples, corresponding to 7.5% of the predictions in the data slice).
Underconfidence	medium	City_Code_Patient == 7.000	Overconfidence rate = 0.073	—	+16.37% than global	For records in your dataset where City_Code_Patient == 7.000, we found a significantly higher number of underconfident predictions (351 samples, corresponding to 7.3% of the predictions in the data slice).
Underconfidence	medium	Ward_Type == "S"	Overconfidence rate = 0.072	—	+15.08% than global	For records in your dataset where Ward_Type == "S", we found a significantly higher number of underconfident predictions (1130 samples, corresponding to 7.2% of the predictions in the data slice).
Underconfidence	medium	Hospital_region_code == "Z"	Overconfidence rate = 0.071	—	+12.94% than global	For records in your dataset where Hospital_region_code == "Z", we found a significantly higher number of underconfident predictions (889 samples, corresponding to 7.1% of the predictions in the data slice).
Underconfidence	medium	Hospital_type_code == "c"	Overconfidence rate = 0.070	—	+11.68% than global	For records in your dataset where Hospital_type_code == "c", we found a significantly higher number of underconfident predictions (644 samples, corresponding to 7.0% of the predictions in the data slice).
Underconfidence	medium	City_Code_Hospital == 2	Overconfidence rate = 0.070	—	+10.96% than global	For records in your dataset where City_Code_Hospital == 2, we found a significantly higher number of underconfident predictions (722 samples, corresponding to 7.0% of the predictions in the data slice).
Underconfidence	medium	Ward_Facility_Code == "D"	Overconfidence rate = 0.070	—	+10.96% than global	For records in your dataset where Ward_Facility_Code == "D", we found a significantly higher number of underconfident predictions (722 samples, corresponding to 7.0% of the predictions in the data slice).

View full report

Debug your scan issues (demo)