diff options
| -rw-r--r-- | _posts/2012-05-02-use-case-table-for-fminer.md | 7 |
1 files changed, 3 insertions, 4 deletions
diff --git a/_posts/2012-05-02-use-case-table-for-fminer.md b/_posts/2012-05-02-use-case-table-for-fminer.md index 4468e5b..791ed2d 100644 --- a/_posts/2012-05-02-use-case-table-for-fminer.md +++ b/_posts/2012-05-02-use-case-table-for-fminer.md @@ -17,7 +17,7 @@ tags: [Algorithm, Fminer, Feature Generation, BBRC, LAST-PM, Tutorials] |Usecase |Commandline |Webservice |API| |:-------|:-----------|:----------|:--| |_n_-Frequent Subgraph Mining (w/o adding activities)|-f _n_|-|SetMinfreq( _n_ )<br />SetChisqActive(false)| -|_n_-Frequent Subgraph Mining (w added activities)|-d -b -f _n_ -p _0.0|-|SetDynamicUpperBound(false)<br /> SetBackbone(false)<br /> SetMinfreq( _n_ )<br /> SetChisqSig( _0.0_ )| +|_n_-Frequent Subgraph Mining (w added activities)|-d -b -f _n_ -p _0.0|backbone=false<br /> min_frequency= _n_<br /> min_chisq_significance= 0|SetDynamicUpperBound(false)<br /> SetBackbone(false)<br /> SetMinfreq( _n_ )<br /> SetChisqSig( _0.0_ )| |_n_-Frequent<br /> Class-Correlated ( _p_ -value as lower bound) Subgraph Mining|-d -b -f _n_ -p _p_|backbone=false<br /> min_frequency= _n_<br /> min_chisq_significance= _p_ |SetDynamicUpperBound(false)<br /> SetBackbone(false)<br /> SetMinfreq( _n_ )<br /> SetChisqSig( _p_ )| |Backbone Refinement Class Mining ( _p_ -value as lower bound)|-f _n_ -p _p_ |min_frequency= _n_<br /> min_chisq_significance= _p_|SetMinfreq( _n_ )<br /> SetChisqSig( _p_ )| @@ -35,10 +35,9 @@ tags: [Algorithm, Fminer, Feature Generation, BBRC, LAST-PM, Tutorials] **Note:** - -* In BBRC, two modes for _n_-frequent subgraph mining are available. The first one does not use activity information, whereas the second one uses it to indicate occurrences of subgraphs per class (for nominal activities), which can be interesting despite deliberately not taking class correlation into account during the mining process. - +* In BBRC, two modes for _n_-frequent subgraph mining are available. The first one does not use activity information, whereas the second one uses it to indicate occurrences of subgraphs per class (for nominal activities), which can be interesting despite deliberately not taking class correlation into account during the mining process. Importantly, if you use the webservice for this setting, make sure that every compound _has_ an activity, otherwise it will be ignored. + * For all settings (except web service) [environment variables](https://raw.github.com/amaunz/fminer2/master/fminer/README) play a major role for the output format. Supported are gSpan and SMARTS, the latter as CSV or YAML. Moreover, _p_-values can be requested instead of $latex \chi^2$ test statistics.** **A special case is the [output of hit counts](/algorithm/2012/05/02/support-calculation-in-fminer). While on the command line, there is a special environment variable, the webservice must be passed a special switch: -d "nr_hits". |