Up One Level Febrl - Freely extensible biomedical record linkage Contents Index
Previous: Up: 9.2 Field Comparison Functions Next:

9.2.7 Numeric Comparison with Percentage Tolerance  'FieldComparatorNumericPerc'

This field comparator is for numeric fields, where a given maximal percentage difference can be tolerated. This percentage has to be set by the argument max_perc_diff as a value between $0.0$ and $100.0$. The default value is $0.0$, i.e. no difference is tolerated.

The agreement weight is returned if the numbers are the same, and the disagreement weight if the percentage difference is larger than the max_perc_diff value. If the percentage difference between the two values is larger than $0.0$ but equal to or smaller than max_perc_diff the partial agreement weight is calculated according to the following formula.

\begin{eqnarray*} partial\_agreement & = & agree\_weight - \left( \frac{perc\_di... ...f+1} \right) * \\ ~& ~& (agree\_weight + abs(disagree\_weight)) \end{eqnarray*}


where the percentage difference between the two values is calculated as

\begin{displaymath} perc\_diff = \left( \frac{abs(value\_A - value\_B)} {max(abs(value\_A),abs(value\_B)} \right) * 100.0 \end{displaymath}

In its current version no frequency dependent weight calculation is possible, thus the arguments frequency_table, freq_table_max_weight and freq_table_min_weight can not be used with this field comparator function.

Up One Level Febrl - Freely extensible biomedical record linkage Contents Index
Previous: Up: 9.2 Field Comparison Functions Next:
Release 0.3.1, documentation updated on July 1, 2005.