Table 2.
The weights used to generate error bars for each assessed data sample based upon the posteriors of the quality states. The weight w1 corresponds to the sensor accuracies specified by the manufacturer whilst the other values were based upon the empirical values in [14] but modified to incorporate 4 classes as opposed to 3 classes.
| Sensed Phenomenon | w1 - Good | w2 - Prob Good | w3 - Prob Bad | w4 - Bad |
|---|---|---|---|---|
| Conductivity | 0.1 | 0.3 | 1.2 | 3 |
| Temperature | 0.2 | 1 | 4 | 8 |