diff options
| author | Birte Kristina Friesel <birte.friesel@uos.de> | 2024-01-11 12:12:26 +0100 |
|---|---|---|
| committer | Birte Kristina Friesel <birte.friesel@uos.de> | 2024-01-11 12:12:26 +0100 |
| commit | a91a70e1d0f420a981b9e33fa11c424212a42b52 (patch) | |
| tree | 1d518472db95c00c430ede81c908904723a73585 | |
| parent | 325bd733f2d1af8ed861d865d24cbad98b552007 (diff) | |
doc: mention that --plot-param also shows model predictions
| -rw-r--r-- | README.md | 2 | ||||
| -rw-r--r-- | doc/analysis-visual.md | 3 | ||||
| -rw-r--r-- | doc/model-visual.md | 5 | ||||
| -rw-r--r-- | doc/modeling-method.md | 2 |
4 files changed, 11 insertions, 1 deletions
@@ -135,7 +135,7 @@ The following variables may be set to alter the behaviour of dfatool components. | `DFATOOL_DTREE_IGNORE_IRRELEVANT_PARAMS` | 0, **1** | Ignore parameters deemed irrelevant by stddev heuristic during regression tree generation | | `DFATOOL_DTREE_LOSS_IGNORE_SCALAR` | **0**, 1 | Ignore scalar parameters when computing the loss for split node candidates. Instead of computing the loss of a single partition for each `x_i == j`, compute the loss of partitions for `x_i == j` in which non-scalar parameters vary and scalar parameters are constant. This way, scalar parameters do not affect the decision about which non-scalar parameter to use for splitting. | | `DFATOOL_PARAM_CATEGORIAL_TO_SCALAR` | **0**, 1 | Some models (e.g. FOL, sklearn CART, XGBoost) do not support categorial parameters. Ignore them (0) or convert them to scalar indexes (1). | -| `DFATOOL_FIT_FOL` | **0**, 1 | Build a first-order linear function (i.e., a * param1 + b * param2 + ...) instead of more complex functions or tree structures. | +| `DFATOOL_FIT_FOL` | **0**, 1 | Build a first-order linear function (i.e., a * param1 + b * param2 + ...) instead of more complex functions or tree structures. Must not be combined with `--force-tree`. | | `DFATOOL_FOL_SECOND_ORDER` || **0**, 1 | Add second-order components (interaction of feature pairs) to first-order linear function. | ## Examples diff --git a/doc/analysis-visual.md b/doc/analysis-visual.md index dad5bb6..b069bd0 100644 --- a/doc/analysis-visual.md +++ b/doc/analysis-visual.md @@ -22,4 +22,7 @@ parameter name. It shows each distinct configuration (parameter/NFP value) in a different colour. Combining it with `--filter-param` and `--ignore-param` may help de-clutter the plot. +dfatool will additionally plot the predicted performance for each distinct +configuration as a solid line. +  diff --git a/doc/model-visual.md b/doc/model-visual.md index 7830421..564cf3e 100644 --- a/doc/model-visual.md +++ b/doc/model-visual.md @@ -16,3 +16,8 @@ accessible via `dot -Tpng filename.dot | feh -`. In case of regression forests (XGBoost), dfatool exports the individual trees to `PREFIX(name)-(attribute).(index).dot`. + +## Plotting Model Predictions for Individual Configurations + +`--plot-param name:attribute:label` displays both raw readings (as points, see +[[analysis-visual.md]]) and the corresponding performance model (as lines). diff --git a/doc/modeling-method.md b/doc/modeling-method.md index d29ab3a..f369cc0 100644 --- a/doc/modeling-method.md +++ b/doc/modeling-method.md @@ -39,6 +39,8 @@ All of these are valid regression model trees. * `--force-tree` builds a tree structure even if dfatool's heuristic indicates that no non-integer parameter affects the modeled performance attribute. * `DFATOOL_DTREE_IGNORE_IRRELEVANT_PARAMS=0` disables the relevant parameter detection heuristic when building the tree structure. By default, irrelevant parameters cannot end up as decision nodes. * `DFATOOL_FIT_LINEAR_ONLY=1` makes RMT behave more like LMT by only considering linear functions in leaf nodes. +* `DFATOOL_FIT_FOL=1` +* `DFATOOL_PARAM_CATEGORIAL_TO_SCALAR=1` * `DFATOOL_SKIP_CODEPENDENT_CHECK=1` * `DFATOOL_REGRESSION_SAFE_FUNCTIONS=1` |