Data Audit Report
- Raw shape:
353 rows x 13 columns
- Complete-case modeling shape:
346 rows x 13 columns
- Duplicate rows after basic cleaning:
10
Missing Values
| index |
value |
| Age |
1 |
| Machine_Type |
1 |
| Pressure |
1 |
| Maintenance |
1 |
| Temperature |
1 |
| Engine_problem |
0 |
| Run_Hours |
1 |
| Sensor1 |
0 |
| Sensor2 |
0 |
| Sensor3 |
0 |
| Sensor4 |
0 |
| Sensor5 |
1 |
| Operation_modes |
0 |
Class Counts
The current label assumption is 0 = Failure, 1 = Production.
| Operation_modes |
count |
| Failure |
253 |
| Production |
93 |
Feature-Target Correlation
| index |
Operation_modes |
| Sensor5 |
-0.809 |
| Sensor2 |
-0.26 |
| Run_Hours |
-0.172 |
| Temperature |
-0.154 |
| Sensor3 |
-0.076 |
| Sensor4 |
-0.001 |
| Machine_Type |
0.032 |
| Maintenance |
0.04 |
| Engine_problem |
0.085 |
| Pressure |
0.121 |
| Sensor1 |
0.203 |
| Age |
0.317 |
Sensor5 By Class
| Operation_modes |
count |
mean |
std |
min |
25% |
50% |
75% |
max |
| 0.0 |
253.0 |
170.826 |
52.903 |
34.0 |
130.0 |
160.0 |
213.0 |
333.0 |
| 1.0 |
93.0 |
24.226 |
27.213 |
8.0 |
12.0 |
12.0 |
31.0 |
150.0 |
Outlier Notes
| index |
min |
max |
iqr_lower |
iqr_upper |
outside_iqr_rule |
| Pressure |
23.0 |
7861.0 |
-652.375 |
1322.625 |
28.0 |
| Run_Hours |
25100.0 |
850000.0 |
57500.0 |
491500.0 |
21.0 |
| Sensor1 |
0.5 |
9.4 |
-0.35 |
3.25 |
13.0 |
| Sensor2 |
113.0 |
232.0 |
124.125 |
151.125 |
51.0 |
| Sensor5 |
8.0 |
333.0 |
-168.0 |
416.0 |
0.0 |
Warnings
- README class-distribution language should be reconciled with observed label counts.
- Sensor5 is highly correlated with the target and must be checked for leakage or status-code encoding.
- Random row splits are not sufficient for predictive maintenance validation when timestamps or asset IDs exist.