diff options
Diffstat (limited to 'src/bm/index.cpp')
-rw-r--r-- | src/bm/index.cpp | 31 |
1 files changed, 31 insertions, 0 deletions
diff --git a/src/bm/index.cpp b/src/bm/index.cpp index 6bc6c78..be29787 100644 --- a/src/bm/index.cpp +++ b/src/bm/index.cpp @@ -280,5 +280,36 @@ bool IndexAlgorithm1::computeValues( } } + + const bool computeDataQualityStatistics = false; // 4 NOW + if (computeDataQualityStatistics) { + // Step 1: Compute the MaxESSTotalCount and MaxESSStableCount for each + // contributing RH (compute for the exact median-smoothed + // values that formed the basis for computing the index) + // + // Step 2: Compute the complete distribution of MaxESSTotalCount + // (note that the number of distinct counts are likely to be only a small + // fraction of the number of RHs): + // + // TC(c1) = <# of RHs with a MaxESSTotalCount of c1> + // TC(c2) = <# of RHs with a MaxESSTotalCount of c2> + // ... + // TC(cN) = <# of RHs with a MaxESSTotalCount of cN> + // + // (TC = Total Count, and N is number of distinct counts) + // + // Step 3: Compute the percentile distribution (for the 10 levels 10%, 20%, ..., 100%) + // of the SF (stability fraction) values (where the SF for a given + // RH is MaxESSStableCount / MaxESSTotalCount): + // + // SFP(100) = <the max SF value for the worst 100% of the RHs (i.e. all RHs!)> + // SFP(90) = <the max SF value for the worst 90% of the RHs> + // SFP(80) = <the max SF value for the worst 80% of the RHs> + // ... + // SFP(10) = <the max SF value for the worst 10% of the RHs> + // + // (SFP = Stability Fraction Percentile) + } + return true; } |