Commit 135449e4 authored by Tomke Schroer's avatar Tomke Schroer
Browse files
parents 398cf270 197545d8
Pipeline #3961508 failed with stages
in 29 minutes and 4 seconds
......@@ -7,6 +7,7 @@
- test_plotting_umami_dips
- test_plotting_umami_dl1
- test_plotting_umami_umami
- test_preprocessing_umami_importance_no_replace
- unittest_parallel
- test_examples
......
......@@ -63,9 +63,8 @@ sphinx-docs:
stage: prepare_docs
image: '${CI_REGISTRY}/${CI_PROJECT_NAMESPACE}/umami/$IMAGE_TYPE'
script:
- pip install ipython
- pip install Sphinx
- pip install sphinx_rtd_theme
- pip install pydata-sphinx-theme
- source run_setup.sh
- cd docs/sphinx
- mkdir source
......
......@@ -65,6 +65,19 @@ test_preprocessing_dl1r_count:
- test_preprocessing_dl1r/
- coverage_files/
test_preprocessing_umami_importance_no_replace:
<<: *test_template
stage: integration_test_preprocessing
script:
- pytest --cov=./ --cov-report= ./umami/tests/integration/test_preprocessing.py -k "test_preprocessing_umami_importance_no_replace" -v -s --junitxml=report.xml
- cp .coverage ./coverage_files/.coverage.test_preprocessing_umami_importance_no_replace
artifacts:
<<: *artifact_template
paths:
- plots/
- test_preprocessing_umami_importance_no_replace/
- coverage_files/
test_preprocessing_umami_count:
<<: *test_template
stage: integration_test_preprocessing
......
......@@ -3,9 +3,12 @@
### Latest
- Fix for the "exclude" funtionality [!528](https://gitlab.cern.ch/atlas-flavor-tagging-tools/algorithms/umami/-/merge_requests/528)
- Adding Plotting API to PlottingFunctions in the eval tools [!532](https://gitlab.cern.ch/atlas-flavor-tagging-tools/algorithms/umami/-/merge_requests/532)
- Fix for the "exclude" funtionality [!528](https://gitlab.cern.ch/atlas-flavor-tagging-tools/algorithms/umami/-/merge_requests/528)
- Adding metrics to Callback functions + Fixing model summary issue [!526](https://gitlab.cern.ch/atlas-flavor-tagging-tools/algorithms/umami/-/merge_requests/526)
- Improved compression settings during scaling and writing [!527](https://gitlab.cern.ch/atlas-flavor-tagging-tools/algorithms/umami/-/merge_requests/527)
- Add documentation and integration tests for importance sampling without replacement method [!502](https://gitlab.cern.ch/atlas-flavor-tagging-tools/algorithms/umami/-/merge_requests/502)
- (Plotting API) Update training plots to plotting API [!515](https://gitlab.cern.ch/atlas-flavor-tagging-tools/algorithms/umami/-/merge_requests/515)
- Fix validation values json in continue_training [!516](https://gitlab.cern.ch/atlas-flavor-tagging-tools/algorithms/umami/-/merge_requests/516/)
- Fixing bunch of invalid-name pylint errors [!522](https://gitlab.cern.ch/atlas-flavor-tagging-tools/algorithms/umami/-/merge_requests/522)
......
# Input Correlations API
Correlations between input variables can be made visible with the `input_correlations.py` script that can be found [here](https://gitlab.cern.ch/atlas-flavor-tagging-tools/algorithms/umami/-/blob/master/examples/input_correlations.py). It plots a linear correlation matrix and scatterplots between all variables given by a yaml variable file.
Correlations between input variables can be made visible with the `input_correlations.py` script that can be found [here](https://gitlab.cern.ch/atlas-flavor-tagging-tools/algorithms/umami/-/blob/master/examples/plotting/input_correlations.py). It plots a linear correlation matrix and scatterplots between all variables given by a yaml variable file.
???+ example "Correlation Matrix"
![input_correlations](../../ci_assets/correlation_matrix.png)
......
# Pie charts plotting API
In the following a small example how to plot a pie chart with the umami python api.
To set up the inputs for the plots, have a look [here](./index.md).
Then we can start the actual plotting part
???+ example "Pie charts plot code"
![pies](../../ci_assets/pie_chart_HadronConeExclTruthLabelID.png)
```py linenums="1"
§§§examples/plotting/plot_pie.py§§§
```
\ No newline at end of file
......@@ -249,7 +249,7 @@ sampling:
```
In `sampling`, we can define the method which is used in the preprocessing for resampling. `method` defines the method which is used. Currently available are `count`, `pdf` and `weighting`. The details of the different sampling methods are explained at their respective sections. The here shown config is for the `count` method.
In `sampling`, we can define the method which is used in the preprocessing for resampling. `method` defines the method which is used. Currently available are `count`, `pdf`, `importance_no_replace` and `weighting`. The details of the different sampling methods are explained at their respective sections. The here shown config is for the `count` method.
An important part are the `class_labels` which are defined here. You can define which flavours are used in the preprocessing. The name of the available flavours can be find [here](https://gitlab.cern.ch/atlas-flavor-tagging-tools/algorithms/umami/-/blob/master/umami/configs/global_config.yaml). Add the names of those to the list to add them to the preprocessing. **PLEASE KEEP THE ORDERING CONSTANT! THIS IS VERY IMPORTANT**. This list must be the same as the one in the train config!
......@@ -389,7 +389,7 @@ Standard undersampling approach. Undersamples all flavours to the statistically
| `custom_njets_initial` | `dict` | Used jets per sample to ensure a smooth hybrid sample of ttbar and zprime, we need to define some empirically derived values for the ttbar samples. |
| `samples` | `dict` | You need to define them for `ttbar` and `zprime`. The samples defined in here are the ones we prepared in the step above. To ensure a smooth hybrid sample of ttbar and zprime, we need to define some empirically derived values for the ttbar samples in `custom_njets_initial`. |
#### PDF Sampling
#### Importance Sampling With Replacement (PDF Sampling)
The PDF sampling method is based on the principles of importance sampling. If your sample's statistics are small and/or your lowest distribution is other than the target distribution (in case of b-tagging, this is the b-jet distribution), you can force the b-jet distribution shape on the other jet flavour distributions. This will ensure all the distributions have the target distribution shape and the same fractions for the two given resampling variables. To enforce the same shape and number of jets per `pT` and _η_ bin, the statistically higher flavours are undersampled and the statistically lower flavours are upsampled to the target flavour. An example for the reprocessing config file which uses the pdf sampling can be found [here](https://gitlab.cern.ch/atlas-flavor-tagging-tools/algorithms/umami/-/blob/master/examples/PFlow-Preprocessing-taus.yaml). In this case, four different flavours are used.
......@@ -437,6 +437,52 @@ First are the bins for the two resampling variables. You need to define a nested
| `samples` | `dict` | You need to define them for `ttbar` and `zprime`. The samples defined in here are the ones we prepared in the step above. To ensure a smooth hybrid sample of ttbar and zprime, we need to define some empirically derived values for the ttbar samples in `custom_njets_initial`. |
| `max_upsampling_ratio` | `dict` | Here you can define for the different samples, which are defined in the `samples` section, a maximal ratio of upsampling. If there are not enough cjets and the `max_upsampling_ratio` is reached, the form of the distribution is applied but not the number. So there can be different numbers of jets per bin per class, but the shape of distributions will still be the same (if you normalise them). |
#### Importance Sampling Without Replacement
Method based on the principles of importance sampling. This method is similar to the count method but with the added flexibility of being able to take a target distribution which all the other distributions should fall under. The implementation also ensuring same fractions per flavor. The difference between this method and the PDF sampling method, is that examples/events are not repeated. You can force the b-jet distribution shape on the other jet flavour distributions by specifying the target distribution to be the b-jets. This will ensure all the distributions have the b-jets shape and same fractions for the two given resampling variables `pT` and _η_ . To enforce the same shape and number of jets per `pT` and _η_ bin, first the sampling probabilityies are calculated using `target / distribution_i`, where `distribution_i` is for each flavour, then the distributions are scaled up/down using the maximum sampling probability. The statistically higher flavours are undersampled and the statistically lower flavours are first scaled then downsampled to the target flavour.
The options for the this method are similar to the ones from the `count` method.
```yaml
sampling:
# Downsampling method that gives same fractions and shape
# distributions given a target distribution, here the b-jets
method: importance_no_replace
options:
# Specify the target distribution
target_distribution: bjets
# jet variables used
sampling_variables:
- pt_btagJes:
# bins take either a list containing the np.linspace arguments
# or a list of them
bins: bins: [0, 15e5, 250]
- absEta_btagJes:
bins: [0, 2.5, 9]
# Decide, which of the in preparation defined samples are used in the resampling.
samples:
ttbar:
- training_ttbar_bjets
- training_ttbar_cjets
- training_ttbar_ujets
zprime:
- training_zprime_bjets
- training_zprime_cjets
- training_zprime_ujets
# Set to -1 or don't include this to use all the available jets
njets: -1
```
| Setting | Type | Explanation |
| ------- | ---- | ----------- |
| `sampling_variables` | `list` | Needs exactly 2 variables. Sampling variables which are used for resampling. The example shows this for the `pt_btagJes` and `absEta_btagJes` variables. In case of the `pdf` method, you define a nested list (one sublist for each category (ttbar or zprime)) with the first and last bin edge and the number of bins to use (np.linespace arguments). |
| `samples` | `dict` | Needs all the different samples for `ttbar` and `zprime`. The samples defined in here are the ones we prepared in the step above.|
| `target_distribution` | `str` | Target distribution to be used for computing the sampling probabilities relative to. This ensures all the final resampled distributions have the same shape and fraction as the target distribution. Default is the `bjets`. |
#### Weighting Sampling
Alternatively you can calculate weights between the flavor of bins in the 2d(pt,eta) histogram and write out all jets. These weights can be forwarded to the training to weigh the loss function of the training. If you want to use them don't forget to set `bool_attach_sample_weights` to `True`.
......
......@@ -65,7 +65,7 @@ language = None
exclude_patterns = ["_build", "Thumbs.db", ".DS_Store"]
# The name of the Pygments (syntax highlighting) style to use.
pygments_style = None
pygments_style = "friendly"
# -- Options for HTML output -------------------------------------------------
......@@ -73,13 +73,22 @@ pygments_style = None
# The theme to use for HTML and HTML Help pages. See the documentation for
# a list of builtin themes.
#
html_theme = "sphinx_rtd_theme"
html_theme = "pydata_sphinx_theme"
# Theme options are theme-specific and customize the look and feel of a theme
# further. For a list of options available for each theme, see the
# documentation.
#
# html_theme_options = {}
html_theme_options = {
"icon_links": [
{
"name": "GitLab",
"url": "https://gitlab.cern.ch/atlas-flavor-tagging-tools/algorithms/umami",
"icon": "fab fa-gitlab",
"type": "fontawesome",
},
],
}
# Add any paths that contain custom static files (such as style sheets) here,
# relative to this directory. They are copied after the builtin static files,
......
......@@ -6,10 +6,12 @@
Welcome to umami's documentation!
=================================
.. table of contents in the header of the site
.. toctree::
:maxdepth: 2
:caption: Contents:
:hidden:
Home <self>
modules
......
......@@ -134,7 +134,7 @@ Validation_metrics_settings:
# Plotting API parameters
# fc_value and WP_b are autmoatically added to the plot label
atlas_first_tag: "Internal Simulation"
atlas_first_tag: "Simulation Internal"
atlas_second_tag: "\n$\\sqrt{s}=13$ TeV, PFlow jets"
# Set the datatype of the plots
......
......@@ -114,7 +114,7 @@ Validation_metrics_settings:
# Plotting API parameters
# fc_value and WP_b are autmoatically added to the plot label
atlas_first_tag: "Internal Simulation"
atlas_first_tag: "Simulation Internal"
atlas_second_tag: "\n$\\sqrt{s}=13$ TeV, PFlow jets"
# Set the datatype of the plots
......
......@@ -116,7 +116,7 @@ Validation_metrics_settings:
# Plotting API parameters
# fc_value and WP_b are autmoatically added to the plot label
atlas_first_tag: "Internal Simulation"
atlas_first_tag: "Simulation Internal"
atlas_second_tag: "\n$\\sqrt{s}=13$ TeV, PFlow jets"
# Set the datatype of the plots
......
......@@ -24,7 +24,7 @@ plot_histo = histogram_plot(
bins=50, # you can also define an integer number for the number of bins
bins_range=(1.1, 4), # only considered if bins is an integer
norm=False,
atlas_first_tag="Internal Simulation",
atlas_first_tag="Simulation Internal",
atlas_second_tag="Example plot for plotting python API",
figsize=(6, 5),
)
......
......@@ -49,7 +49,7 @@ plot_histo = histogram_plot(
ymax_ratio_1=1.5,
ymin_ratio_1=0.5,
atlas_second_tag=(
"$\\sqrt{s}=13$ TeV, PFlow Jets, \n$t\\bar{t}$ Test Sample, $f_{c}=0.018$"
"$\\sqrt{s}=13$ TeV, PFlow jets, \n$t\\bar{t}$ test sample, $f_{c}=0.018$"
),
)
......
"""Example of pie chart plot of the HadronConeExclTruthLabelID"""
from umami.plotting import histogram, histogram_plot
from umami.plotting.utils import get_dummy_2_taggers
# The line below generates dummy data which is similar to a NN output
df = get_dummy_2_taggers(size=12)
HadrTruthLabel_vals = [0, 4, 5, 15]
HadrTruthLabel_labels = ["light-flavour jets", "c-jets", "b-jets", "tau-jets"]
title = "HadronConeExclTruthLabelID"
# the number of bins should be the number of bins needed to have a separat bin
# for every discrete value.
bins = 16
bins_range = (0, 16)
plot_pie = histogram_plot(
n_ratio_panels=0,
title=title,
discrete_vals=HadrTruthLabel_vals,
atlas_second_tag="$\\sqrt{s}=13$ TeV, PFlow Jets,\n$t\\bar{t}$ Test Sample",
bins=bins,
bins_range=bins_range,
draw_errors=False,
vertical_split=True,
plot_pie=True,
pie_colours=None,
pie_labels=HadrTruthLabel_labels,
)
plot_pie.add(histogram(df["HadronConeExclTruthLabelID"]))
plot_pie.draw()
plot_pie.savefig("pie_chart_HadronConeExclTruthLabelID.png")
......@@ -99,7 +99,7 @@ plot_bkg_rej = var_vs_eff_plot(
xlabel=r"$p_{T}$ [GeV]",
logy=False,
atlas_second_tag=(
"$\\sqrt{s}=13$ TeV, PFlow Jets, \n$t\\bar{t}$ Test Sample, $f_{c}=0.018$"
"$\\sqrt{s}=13$ TeV, PFlow jets, \n$t\\bar{t}$ test sample, $f_{c}=0.018$"
),
figsize=(6, 4.5),
)
......@@ -115,7 +115,7 @@ plot_sig_eff = var_vs_eff_plot(
xlabel=r"$p_{T}$ [GeV]",
logy=False,
atlas_second_tag=(
"$\\sqrt{s}=13$ TeV, PFlow Jets, \n$t\\bar{t}$ Test Sample, $f_{c}=0.018$"
"$\\sqrt{s}=13$ TeV, PFlow jets, \n$t\\bar{t}$ test sample, $f_{c}=0.018$"
),
figsize=(6, 4.5),
)
......
......@@ -78,7 +78,7 @@ plot_roc = roc_plot(
ylabel="background rejection",
xlabel="b-jets efficiency",
atlas_second_tag=(
"$\\sqrt{s}=13$ TeV, PFlow Jets, \n$t\\bar{t}$ Test Sample, $f_{c}=0.018$"
"$\\sqrt{s}=13$ TeV, PFlow jets, \n$t\\bar{t}$ test sample, $f_{c}=0.018$"
),
)
plot_roc.add_roc(
......
.default_plot_settings: &default_plot_settings
logy: True
use_atlas_tag: True
atlas_first_tag: "Internal Simulation"
atlas_second_tag: "$\\sqrt{s}$ = 13 TeV, $t\\bar{t}$ PFlow Jets \n30000 Jets"
atlas_first_tag: "Simulation Internal"
atlas_second_tag: "$\\sqrt{s}$ = 13 TeV, $t\\bar{t}$ PFlow jets \n30000 jets"
y_scale: 2
figsize: [7, 5]
......
......@@ -20,23 +20,24 @@ contour_fraction_ttbar:
cjets: 0.1
ujets: 0.9
plot_settings:
yAxisIncrease: 1.3 # Increasing of the y axis so the plots dont collide with labels (mainly AtlasTag)
UseAtlasTag: True # Enable/Disable AtlasTag
AtlasTag: "Internal Simulation"
SecondTag: "\n$\\sqrt{s}=13$ TeV, PFlow Jets,\n$t\\bar{t}$ Test Sample, WP = 77 %"
yAxisAtlasTag: 0.9 # y axis value (1 is top) for atlas tag
y_scale: 1.3 # Increasing of the y axis so the plots dont collide with labels (mainly atlas_first_tag)
use_atlas_tag: True # Enable/Disable atlas_first_tag
atlas_first_tag: "Simulation Internal"
atlas_second_tag: "$\\sqrt{s}=13$ TeV, PFlow jets,\n$t\\bar{t}$ test sample, WP = 77 %"
scores_DL1r: # Each item on this level defines one plot. The name of this key is later used for the name of the output file.
type: "scores"
data_set_name: "ttbar_r21" # data set to use. This is the dict entry name of the file you want to plot in the train config
tagger_name: "DL1" # Name of the tagger: Example: dips_pb -> Name of tagger: dips
class_labels: ["ujets", "cjets", "bjets"] # Classes which are used
main_class: "bjets" # Main class
main_class: "bjets"
models_to_plot:
DL1r:
data_set_name: "ttbar_r21"
tagger_name: "DL1"
class_labels: ["ujets", "cjets", "bjets"]
label: "$t\\bar{t}$"
plot_settings: # All options of the score plot can be changed here
UseAtlasTag: True # Enable/Disable AtlasTag
AtlasTag: "Internal Simulation"
SecondTag: "\n$\\sqrt{s}=13$ TeV, PFlow Jets,\n$t\\bar{t}$ Test Sample"
yAxisAtlasTag: 0.9 # y axis value (1 is top) for atlas tag
use_atlas_tag: True # Enable/Disable atlas_first_tag
atlas_first_tag: "Simulation Internal"
atlas_second_tag: "$\\sqrt{s}=13$ TeV, PFlow jets,\n$t\\bar{t}$ test sample"
confusion_matrix_DL1r:
type: "confusion_matrix"
......@@ -61,82 +62,6 @@ DL1r_light_flavour:
ymax: 1000000
figsize: [7, 6] # [width, hight]
WorkingPoints: [0.60, 0.70, 0.77, 0.85]
UseAtlasTag: True # Enable/Disable AtlasTag
AtlasTag: "Internal Simulation"
SecondTag: "\n$\\sqrt{s}=13$ TeV, PFlow Jets,\n$t\\bar{t}$ Validation Sample, fc=0.018"
yAxisAtlasTag: 0.9 # y axis value (1 is top) for atlas tag
eff_vs_pt:
type: "ROCvsVar"
tagger_name: "DL1"
recompute: True
evaluation_file:
data_set_name: "ttbar_r21"
data_set_for_cut_name: "ttbar_r21"
class_labels: ["ujets", "cjets", "bjets"]
flat_eff: True
efficiency: 70
frac_values: {
"cjets": 0.018,
"ujets": 0.882,
"taujets": 0.1,
}
main_class: "bjets"
variable: pt
max_variable: 1500000
min_variable: 10000
nbin: 100
var_bins: [20, 30, 40, 50, 75, 100, 150, 250]
xticksval: [20, 50, 100, 150, 200, 250]
xticks: ["", "$50$", "$100$", "$150$", "$200$", "$250$"]
plot_settings:
xlabel: "$p_T$ [GeV]"
minor_ticks_frequency: 10
UseAtlasTag: True
AtlasTag: "Internal"
SecondTag: "$\\sqrt{s}$ = 13 TeV, $t\\bar{t}$"
ThirdTag: "Flat efficiency DL1r"
logy: True
eff_vs_pt_comp:
type: "ROCvsVar_comparison"
tagger_name: "DL1"
recompute: False
models_to_plot:
model1:
data_set_name: "ttbar_r21"
data_set_for_cut_name: "ttbar_r21"
class_labels: ["ujets", "cjets", "bjets"]
label: "first"
model2:
evaluation_file:
data_set_name: "ttbar_r21"
data_set_for_cut_name: "ttbar_r21"
recompute: True
tagger_name: "DL1"
class_labels: ["ujets", "cjets", "bjets"]
label: "second"
cut_value:
flat_eff: False
efficiency: 70
frac_values: {
"cjets": 0.018,
"ujets": 0.882,
"taujets": 0.1,
}
main_class: "bjets"
variable: pt
max_variable: 1500000
min_variable: 10000
nbin: 100
var_bins: [20, 30, 40, 50, 75, 100, 150, 250]
xticksval: [20, 50, 100, 150, 200, 250]
xticks: ["", "$50$", "$100$", "$150$", "$200$", "$250$"]
plot_settings:
xlabel: "$p_T$ [GeV]"
minor_ticks_frequency: 10
UseAtlasTag: True
AtlasTag: "Internal"
SecondTag: "$\\sqrt{s}$ = 13 TeV, $t\\bar{t}$"
ThirdTag: "Flat efficiency DL1r"
logy: True
use_atlas_tag: True # Enable/Disable atlas_first_tag
atlas_first_tag: "Simulation Internal"
atlas_second_tag: "$\\sqrt{s}=13$ TeV, PFlow jets,\n$t\\bar{t}$ Validation Sample, fc=0.018"
......@@ -27,31 +27,32 @@ contour_fraction_ttbar:
label: "Umami"
data_set_name: "ttbar_r21"
plot_settings:
yAxisIncrease: 1.3 # Increasing of the y axis so the plots dont collide with labels (mainly AtlasTag)
UseAtlasTag: True # Enable/Disable AtlasTag
AtlasTag: "Internal Simulation"
SecondTag: "\n$\\sqrt{s}=13$ TeV, PFlow Jets,\n$t\\bar{t}$ Test Sample, WP = 77 %"
yAxisAtlasTag: 0.9 # y axis value (1 is top) for atlas tag
y_scale: 1.3 # Increasing of the y axis so the plots dont collide with labels (mainly atlas_first_tag)
use_atlas_tag: True # Enable/Disable atlas_first_tag
atlas_first_tag: "Simulation Internal"
atlas_second_tag: "$\\sqrt{s}=13$ TeV, PFlow jets,\n$t\\bar{t}$ test sample, WP = 77 %"
# Dips, ttbar
scores_Umami_ttbar:
type: "scores"
data_set_name: "ttbar_r21" # data set to use. This is the dict entry name of the file you want to plot in the train config
tagger_name: "umami" # Name of the tagger: Example: dips_pb -> Name of tagger: dips
class_labels: ["ujets", "cjets", "bjets"] # Classes which are used
main_class: "bjets" # Main class
main_class: "bjets"
models_to_plot:
umami_r21:
data_set_name: "ttbar_r21"
tagger_name: "umami"
class_labels: ["ujets", "cjets", "bjets"]
label: "$t\\bar{t}$"
plot_settings:
WorkingPoints: [0.60, 0.70, 0.77, 0.85] # Set Working Point Lines in plot
nBins: 50 # Number of bins
yAxisIncrease: 1.3 # Increasing of the y axis so the plots dont collide with labels (mainly AtlasTag)
UseAtlasTag: True # Enable/Disable AtlasTag
AtlasTag: "Internal Simulation"
SecondTag: "\n$\\sqrt{s}=13$ TeV, PFlow Jets,\n$t\\bar{t}$ Test Sample"
yAxisAtlasTag: 0.9 # y axis value (1 is top) for atlas tag
working_points: [0.60, 0.70, 0.77, 0.85] # Set Working Point Lines in plot
bins: 50 # Number of bins
y_scale: 1.3 # Increasing of the y axis so the plots dont collide with labels (mainly atlas_first_tag)
use_atlas_tag: True # Enable/Disable atlas_first_tag
atlas_first_tag: "Simulation Internal"
atlas_second_tag: "$\\sqrt{s}=13$ TeV, PFlow jets,\n$t\\bar{t}$ test sample"
# Umami, ttbar
scores_Umami_ttbar_comparison:
type: "scores_comparison"
type: "scores"
main_class: "bjets"
models_to_plot:
umami_r21:
......@@ -65,14 +66,13 @@ scores_Umami_ttbar_comparison:
class_labels: ["ujets", "cjets", "bjets"]
label: "$t\\bar{t} 2$"
plot_settings:
WorkingPoints: [0.60, 0.70, 0.77, 0.85] # Set Working Point Lines in plot
nBins: 50 # Number of bins
yAxisIncrease: 1.4 # Increasing of the y axis so the plots dont collide with labels (mainly AtlasTag)
working_points: [0.60, 0.70, 0.77, 0.85] # Set Working Point Lines in plot
bins: 50 # Number of bins
y_scale: 1.4 # Increasing of the y axis so the plots dont collide with labels (mainly atlas_first_tag)
figsize: [8, 6] # [width, hight]
UseAtlasTag: True # Enable/Disable AtlasTag
AtlasTag: "Internal Simulation"
SecondTag: "\n$\\sqrt{s}=13$ TeV, PFlow Jets"
yAxisAtlasTag: 0.9 # y axis value (1 is top) for atlas tag
use_atlas_tag: True # Enable/Disable atlas_first_tag
atlas_first_tag: "Simulation Internal"
atlas_second_tag: "$\\sqrt{s}=13$ TeV, PFlow jets"
Ratio_Cut: [0.5, 1.5]
confusion_matrix_Umami_ttbar:
......@@ -99,53 +99,53 @@ beff_scan_tagger_umami:
ymax: 1000000
figsize: [7, 6] # [width, hight]
WorkingPoints: [0.60, 0.70, 0.77, 0.85]
UseAtlasTag: True # Enable/Disable AtlasTag
AtlasTag: "Internal Simulation"
SecondTag: "\n$\\sqrt{s}=13$ TeV, PFlow Jets,\n$t\\bar{t}$ Validation Sample, fc=0.018"
yAxisAtlasTag: 0.9 # y axis value (1 is top) for atlas tag
use_atlas_tag: True # Enable/Disable atlas_first_tag
atlas_first_tag: "Simulation Internal"
atlas_second_tag: "$\\sqrt{s}=13$ TeV, PFlow jets,\n$t\\bar{t}$ Validation Sample, fc=0.018"
Umami_prob_pb:
type: "probability"
data_set_name: "ttbar_r21"
tagger_name: "umami"
class_labels: ["ujets", "cjets", "bjets"]
prob_class: "bjets"
models_to_plot:
umami:
data_set_name: "ttbar_r21"
label: "UMAMI"
tagger_name: "umami"
class_labels: ["ujets", "cjets", "bjets"]
plot_settings:
logy: True
nBins: 50
yAxisIncrease: 10
UseAtlasTag: True
AtlasTag: "Internal Simulation"
SecondTag: "\n$\\sqrt{s}=13$ TeV, PFlow Jets,\n$t\\bar{t}$ Test Sample"
yAxisAtlasTag: 0.9 # y axis value (1 is top) for atlas tag
bins: 50
y_scale: 1.3
use_atlas_tag: True
atlas_first_tag: "Simulation Internal"
atlas_second_tag: "$\\sqrt{s}=13$ TeV, PFlow jets,\n$t\\bar{t}$ test sample"
Umami_prob_comparison_pb:
type: "probability_comparison"
type: "probability"
prob_class: "bjets"
models_to_plot:
umami_r22:
umami_r21:
data_set_name: "ttbar_r21"
label: "UMAMI"
tagger_name: "umami"
class_labels: ["ujets", "cjets", "bjets"]
umami_r21:
umami_r22:
data_set_name: "ttbar_r21"
label: "Umami DIPS"
tagger_name: "dips"
class_labels: ["ujets", "cjets", "bjets"]
plot_settings:
nBins: 50
bins: 50
logy: False
yAxisIncrease: 100
y_scale: 1.3
figsize: [8, 6]
UseAtlasTag: True
AtlasTag: "Internal Simulation"
SecondTag: "\n$\\sqrt{s}=13$ TeV, PFlow Jets"
yAxisAtlasTag: 0.9
use_atlas_tag: True
atlas_first_tag: "Simulation Internal"
atlas_second_tag: "$\\sqrt{s}=13$ TeV, PFlow jets"
# Scanning b-eff, comparing Umami and DL1r, ttbar