Table 3 Full datasets
From: Investigating the relationship between time and predictive model maintenance
Dataset | Year | Total instances | Fraud instances | Fraud % |
|---|---|---|---|---|
Part B\(^{\text{a}}\) | 2013 | 915,909 | 403 | 0.044 |
2014 | 950,000 | 285 | 0.030 | |
2015 | 972,222 | 175 | 0.018 | |
2016 | 990,000 | 99 | 0.010 | |
Part D | 2013 | 673,913 | 465 | 0.069 |
2014 | 700,000 | 329 | 0.047 | |
2015 | 722,580 | 224 | 0.031 | |
2016 | 750,000 | 135 | 0.018 | |
DMEPOS | 2013 | 293,636 | 323 | 0.110 |
2014 | 283,823 | 193 | 0.068 | |
2015 | 290,243 | 119 | 0.041 | |
2016 | 288,461 | 75 | 0.026 | |
Combined | 2013 | 254,444 | 229 | 0.090 |
2014 | 252,459 | 154 | 0.061 | |
2015 | 257,142 | 90 | 0.035 | |
2016 | 261,904 | 55 | 0.021 |