6. API reference¶

6.1. Module: wltp.experiment¶

The core that accepts a vehicle-model and wltc-classes, runs the simulation and updates the model with results (downscaled velocity & gears-profile).

[0, 0, 1, 1, 1, 2, 2, ... 1, 0, 0]
 <----   cycle time-steps   ---->

 t:||: 0  1  2  3
---+-------------
g1:|[[ 1, 1, 1, 1, ... 1, 1
g2:|   2, 2, 2, 2, ... 2, 2
g3:|   3, 3, 3, 3, ... 3, 3
g4:|   4, 4, 4, 4, ... 4, 4 ]]

6.2. Module: wltp.model¶

Defines the schema, defaults and validation operations for the data consumed and produced by the Experiment.

The model-instance is managed by pandel.Pandel.

wltp.model._get_model_base()[source]¶

The base model for running a WLTC experiment.

It contains some default values for the experiment (ie the default ‘full-load-curve’ for the vehicles). But note that it this model is not valid - you need to override its attributes.

Returns:	a tree with the default values for the experiment.

wltp.model._get_model_schema(additional_properties=False, for_prevalidation=False)[source]¶

Parameters:	additional_properties (bool) – when False, 4rd-step(validation) will scream on any non-schema property found.
Returns:	The json-schema(dict) for input/output of the WLTC experiment.

wltp.model._get_wltc_data()[source]¶

The WLTC-data required to run an experiment (the class-cycles and their attributes)..

Prefer to access wltc-data through model['wltc_data'].

Returns:	a tree

wltp.model._get_wltc_schema()[source]¶

The json-schema for the WLTC-data required to run a WLTC experiment.

:return :dict:

wltp.model.get_class_part_names(cls_name=None)[source]¶

Parameters:	cls_name (str) – one of ‘class1’, ..., ‘class3b’, if missing, returns all 4 part-names

wltp.model.get_class_parts_limits(cls_name, mdl=None, edges=False)[source]¶

Parses the supplied in wltc_data and extracts the part-limits for the specified class-name.

Parameters:	cls_name (str) – one of ‘class1’, ..., ‘class3b’ mdl – the mdl to parse wltc_data from, if ommited, parses the results of _get_wltc_data() edges – when True, embeds internal limits into (0, len)
Returns:	a list with the part-limits, ie for class-3a these are 3 numbers

wltp.model.get_class_pmr_limits(mdl=None, edges=False)[source]¶

Parses the supplied in wltc_data and extracts the part-limits for the specified class-name.

Parameters:	mdl – the mdl to parse wltc_data from, if omitted, parses the results of _get_wltc_data() edges – when True, embeds internal limits into (0, len)
Returns:	a list with the pmr-limits (2 numbers)

wltp.model.get_model_schema(additional_properties=False, for_prevalidation=False)¶

Parameters:	additional_properties (bool) – when False, 4rd-step(validation) will scream on any non-schema property found.
Returns:	The json-schema(dict) for input/output of the WLTC experiment.

wltp.model.merge(a, b, path=[])[source]¶

‘merges b into a

wltp.model.validate_model(mdl, additional_properties=False, iter_errors=False, validate_wltc_data=False, validate_schema=False)[source]¶

Parameters:	iter_errors (bool) – does not fail, but returns a generator of ValidationErrors

>>> validate_model(None)
Traceback (most recent call last):
jsonschema.exceptions.ValidationError: None is not of type 'object'
...

>>> mdl = _get_model_base()
>>> err_generator = validate_model(mdl, iter_errors=True)
>>> sorted(err_generator, key=hash)
[<ValidationError:
...

>>> mdl = _get_model_base()
>>> mdl["vehicle"].update({
...     "unladen_mass":1230,
...     "test_mass":   1300,
...     "v_max":   195,
...     "p_rated": 110.625,
...     "n_rated": 5450,
...     "n_idle":  950,
...     "n_min":   500,
...     "gear_ratios":[120.5, 75, 50, 43, 33, 28],
...     "resistance_coeffs":[100, 0.5, 0.04],
... })
>>> err_generator = validate_model(mdl, iter_errors=True)
>>> len(list(err_generator))
0

6.3. Module: wltp.pandel¶

A pandas-model is a tree of strings, numbers, sequences, dicts, pandas instances and resolvable URI-references, implemented by Pandel.

class wltp.pandel.ModelOperations[source]¶

Bases: wltp.pandel.ModelOperations

Customization functions for traversing, I/O, and converting self-or-descendant branch (sub)model values.

static __new__(inp=None, out=None, conv=None)[source]¶

Parameters:

inp (list) – the args-list to Pandel._read_branch() out – The args to Pandel._write_branch(), that may be specified either as: an args-list, that will apply for all model data-types (lists, dicts & pandas), a map of type –> args-list, where the None key is the catch-all case, a function returning the args-list for some branch-value, with signature: def get_write_branch_args(branch). conv – The conversion-functions (convertors) for the various model’s data-types. The convertors have signature def convert(branch), and they may be specified either as: a map of (from_type, to_type) –> conversion_func(), where the None key is the catch-all case, a “master-switch” function returning the appropriate convertor depending on the requested conversion. The master-function’s signature is def get_convertor(from_branch, to_branch). The minimum convertors demanded by Pandel are (at least, check the code for more): DataFrame <–> dict Series <–> dict ndarray <–> list

class wltp.pandel.Pandel(curate_funcs=())[source]¶

Bases: builtins.object

Builds, validates and stores a pandas-model, a mergeable stack of JSON-schema abiding trees of strings and numbers, assembled with

• sequences,
• dictionaries,
• pandas.DataFrame,
• pandas.Series, and
• URI-references to other model-trees.

Overview

The making of a model involves, among others, schema-validating, reading subtree-branches from URIs, cloning, converting and merging multiple sub-models in a single unified-model tree, without side-effecting given input. All these happen in 4+1 steps:

              ....................... Model Construction .................
 ------------ :  _______    ___________                                  :
/ top_model /==>|Resolve|->|PreValidate|-+                               :
-----------'  : |___0___|  |_____1_____| |                               :
 ------------ :  _______    ___________  |   _____    ________    ______ :   --------
/ base-model/==>|Resolve|->|PreValidate|-+->|Merge|->|Validate|->|Curate|==>/ model /
-----------'  : |___0___|  |_____1_____|    |_ 2__|  |___3____|  |__4+__|:  -------'
              ............................................................

All steps are executed “lazily” using generators (with yield). Before proceeding to the next step, the previous one must have completed successfully. That way, any ad-hoc code in building-step-5(curation), for instance, will not suffer a horrible death due to badly-formed data.

[TODO] The storing of a model simply involves distributing model parts into different files and/or formats, again without side-effecting the unified-model.

Building model

Here is a detailed description of each building-step:

1. _resolve() and substitute any json-references present in the submodels with content-fragments fetched from the referred URIs. The submodels are cloned first, to avoid side-effecting them.

   Although by default a combination of JSON and CSV files is expected, this can be customized, either by the content in the json-ref, within the model (see below), or as explained below.

   The extended json-refs syntax supported provides for passing arguments into _read_branch() and _write_branch() methods. The syntax is easier to explain by showing what the default _global_cntxt corresponds to, for a DataFrame:

   {
     "$ref": "http://example.com/example.json#/foo/bar",
     "$inp": ["AUTO"],
     "$out": ["CSV", "encoding=UTF-8"]
   }
   

   And here what is required to read and (later) store into a HDF5 local file with a predefined name:

   {
     "$ref": "file://./filename.hdf5",
     "$inp": ["AUTO"],
     "$out": ["HDF5"]
   }
   

   Warning

   Step NOT IMPLEMENTED YET!

2. Loosely _prevalidate() each sub-model separately with json-schema, where any pandas-instances (DataFrames and Series) are left as is. It is the duty of the developer to ensure that the prevalidation-schema is loose enough that it allows for various submodel-forms, prior to merging, to pass.

3. Recursively clone and _merge() sub-models in a single unified-model tree. Branches from sub-models higher in the stack override the respective ones from the sub-models below, recursively. Different object types need to be converted appropriately (ie. merging a dict with a DataFrame results into a DataFrame, so the dictionary has to convert to dataframe).

   The required conversions into pandas classes can be customized as explained below. Series and DataFrames cannot merge together, and Sequences do not merge with any other object-type (themselfs included), they just “overwrite”.

   The default convertor-functions defined both for submodels and models are listed in the following table:

   From:	To:	Method:
   dict	DataFrame	pd.DataFrame (the constructor)
   DataFrame	dict	lambda df: df.to_dict('list')
   dict	Series	pd.Series (the constructor)
   Series	dict	lambda sr: sr.to_dict()

4. Strictly json-_validate() the unified-model (ie enforcing required schema-rules).

   The required conversions from pandas classes can be customized as explained below.

   The default convertor-functions are the same as above.

5. (Optionally) Apply the _curate() functions on the the model to enforce dependencies and/or any ad-hoc generation-rules among the data. You can think of bash-like expansion patterns, like ${/some/path:=$HOME} or expressions like %len(../other/path).

Storing model

When storing model-parts, if unspecified, the filenames to write into will be deduced from the jsonpointer-path of the $out‘s parent, by substituting “strange” chars with undescores(_).

Warning

Functionality NOT IMPLEMENTED YET!

Customization

Some operations within steps (namely conversion and IO) can be customized by the following means (from lower to higher precedance):

1. The global-default ModelOperations instance on the _global_cntxt, applied on both submodels and unified-model.

   For example to channel the whole reading/writing of models through HDF5 data-format, it would suffice to modify the _global_cntxt like that:

   pm = FooPandelModel()                        ## some concrete model-maker
   io_args = ["HDF5"]
   pm.mod_global_operations(inp=io_args, out=io_args)
   

2. [TODO] Extra-properties on the json-schema applied on both submodels and unified-model for the specific path defined. The supported properties are the non-functional properties of ModelOperations.

4. Specific-properties regarding IO operations within each submodel - see the resolve building-step, above.

3. Context-maps of json_paths –> ModelOperations instances, installed by add_submodel() and unified_contexts on the model-maker. They apply to self-or-descedant subtree of each model.

   The json_path is a strings obeying a simplified json-pointer syntax (no char-normalizations yet), ie /some/foo/1/pointer. An empty-string('') matches all model.

   When multiple convertors match for a model-value, the selected convertor to be used is the most specific one (the one with longest prefix). For instance, on the model:

   [ { "foo": { "bar": 0 } } ]
   

   all of the following would match the 0 value:

   • the global-default _global_cntxt,
   • /, and
   • /0/foo

   but only the last’s context-props will be applied.

Atributes

model¶

The model-tree that will receive the merged submodels after build() has been invoked. Depending on the submodels, the top-value can be any of the supported model data-types.

_submodel_tuples¶

The stack of (submodel, path_ops) tuples. The list’s 1st element is the base-model, the last one, the top-model. Use the add_submodel() to build this list.

_global_cntxt¶

A ModelOperations instance acting as the global-default context for the unified-model and all submodels. Use mod_global_operations() to modify it.

_curate_funcs¶

The sequence of curate functions to be executed as the final step by _curate(). They are “normal” functions (not generators) with signature:

def curate_func(model_maker):
    pass      ## ie: modify ``model_maker.model``.

Better specify this list of functions on construction time.

_errored¶

An internal boolean flag that becomes True if any build-step has failed, to halt proceeding to the next one. It is None if build has not started yet.

Examples

The basic usage requires to subclass your own model-maker, just so that a json-schema is provided for both validation-steps, 2 & 4:

>>> from collections import OrderedDict as od                           ## Json is better with stable keys-order

>>> class MyModel(Pandel):
...     def _get_json_schema(self, is_prevalidation):
...         return {                                                    ## Define the json-schema.
...             '$schema': 'http://json-schema.org/draft-04/schema#',
...             'required': [] if is_prevalidation else ['a', 'b'],     ## Prevalidation is more loose.
...             'properties': {
...                 'a': {'type': 'string'},
...                 'b': {'type': 'number'},
...                 'c': {'type': 'number'},
...             }
...         }

Then you can instanciate it and add your submodels:

>>> mm = MyModel()
>>> mm.add_submodel(od(a='foo', b=1))                                   ## submodel-1 (base)
>>> mm.add_submodel(pd.Series(od(a='bar', c=2)))                        ## submodel-2 (top-model)

You then have to build the final unified-model (any validation errors would be reported at this point):

>>> mdl = mm.build()

Note that you can also access the unified-model in the model attribute. You can now interogate it:

>>> mdl['a'] == 'bar'                       ## Value overridden by top-model
True
>>> mdl['b'] == 1                           ## Value left intact from base-model
True
>>> mdl['c'] == 2                           ## New value from top-model
True

Lets try to build with invalid submodels:

>>> mm = MyModel()
>>> mm.add_submodel({'a': 1})               ## According to the schema, this should have been a string,
>>> mm.add_submodel({'b': 'string'})        ## and this one, a number.

>>> sorted(mm.build_iter(), key=lambda ex: ex.message)                   ## Fetch a list with all validation errors.
[<ValidationError: "'string' is not of type 'number'">,
 <ValidationError: "1 is not of type 'string'">,
 <ValidationError: 'Gave-up building model after step 1.prevalidate (out of 4).'>]

>>> mdl = mm.model
>>> mdl is None                                     ## No model constructed, failed before merging.
True

And lets try to build with valid submodels but invalid merged-one:

>>> mm = MyModel()
>>> mm.add_submodel({'a': 'a str'})
>>> mm.add_submodel({'c': 1})

>>> sorted(mm.build_iter(), key=lambda ex: ex.message)        ## Missing required('b') prop rom merged-model.
[<ValidationError: "'b' is a required property">,
 <ValidationError: 'Gave-up building model after step 3.validate (out of 4).'>]

__init__(curate_funcs=())[source]¶

Parameters:	curate_funcs (sequence) – See _curate_funcs.

__metaclass__¶

alias of ABCMeta

_clone_and_merge_submodels(a, b, path='')[source]¶

‘ Recursively merge b into a, cloning both.

_curate()[source]¶

Step-4: Invokes any curate-functions found in _curate_funcs.

_get_json_schema(is_prevalidation)[source]¶

Returns:	a json schema, more loose when prevalidation for each case
Return type:	dictionary

_merge()[source]¶

Step-2

_prevalidate()[source]¶

Step-1

_read_branch()[source]¶

Reads model-branches during resolve step.

_resolve()[source]¶

Step-1

_select_context(path, branch)[source]¶

Finds which context to use while visiting model-nodes, by enforcing the precedance-rules described in the Customizations.

Parameters:	path (str) – the branch’s jsonpointer-path branch (str) – the actual branch’s node
Returns:	the selected ModelOperations

_validate()[source]¶

Step-3

_write_branch()[source]¶

Writes model-branches during distribute step.

add_submodel(model, path_ops=None)[source]¶

Pushes on top a submodel, along with its context-map.

Parameters:	model – the model-tree (sequence, mapping, pandas-types) path_ops (dict) – A map of json_paths –> ModelOperations instances acting on the unified-model. The path_ops may often be empty.

Examples

To change the default DataFrame –> dictionary convertor for a submodel, use the following:

>>> mdl = {'foo': 'bar'}
>>> submdl = ModelOperations(mdl, conv={(pd.DataFrame, dict): lambda df: df.to_dict('record')})

build()[source]¶

Attempts to build the model by exhausting build_iter(), or raises its 1st error.

Use this method when you do not want to waste time getting the full list of errors.

build_iter()[source]¶

Iteratively build model, yielding any problems as ValidationError instances.

For debugging, the unified model at model my contain intermediate results at any time, even if construction has failed. Check the _errored flag if neccessary.

mod_global_operations(operations=None, **cntxt_kwargs)[source]¶

Since it is the fall-back operation for conversions and IO operation, it must exist and have all its props well-defined for the class to work correctly.

Parameters:	operations (ModelOperations) – Replaces values of the installed context with non-empty values from this one. cntxt_kwargs – Replaces the keyworded-values on the existing operations. See ModelOperations for supported keywords.

unified_contexts[source]¶

A map of json_paths –> ModelOperations instances acting on the unified-model.

class wltp.pandel.PandelVisitor(schema, types=(), resolver=None, format_checker=None, skip_meta_validation=False)[source]¶

Bases: jsonschema.validators.Validator

A customized Draft4Validator suporting instance-trees with pandas and numpy objects, natively.

Any pandas or numpy instance (for example obj) is treated like that:

Python Type	JSON Equivalence
pandas.DataFrame	as object json-type, with obj.columns as keys, and obj[col].values as values
pandas.Series	as object json-type, with obj.index as keys, and obj.values as values
np.ndarray, list, tuple	as array json-type

Note that the value of each dataFrame column is a :ndarray instances.

The simplest validations of an object or a pandas-instance is like this:

>>> import pandas as pd

>>> schema = {
...     'type': 'object',
... }
>>> pv = PandelVisitor(schema)

>>> pv.validate({'foo': 'bar'})
>>> pv.validate(pd.Series({'foo': 1}))
>>> pv.validate([1,2])                                       ## A sequence is invalid here.
Traceback (most recent call last):
...
jsonschema.exceptions.ValidationError: [1, 2] is not of type 'object'

Failed validating 'type' in schema:
    {'type': 'object'}

On instance:
    [1, 2]

Or demanding specific properties with required and no additionalProperties:

>>> schema = {
...     'type':     'object',
...     'required': ['foo'],
...    'additionalProperties': False,
...    'properties': {
...        'foo': {}
...    }
... }
>>> pv = PandelVisitor(schema)

>>> pv.validate(pd.Series({'foo': 1}))
>>> pv.validate(pd.Series({'foo': 1, 'bar': 2}))             ## Additional 'bar' is present!
Traceback (most recent call last):
...
jsonschema.exceptions.ValidationError: Additional properties are not allowed ('bar' was unexpected)

Failed validating 'additionalProperties' in schema:
    {'additionalProperties': False,
     'properties': {'foo': {}},
     'required': ['foo'],
     'type': 'object'}

On instance:
    bar    2
    foo    1
    dtype: int64

>>> pv.validate(pd.Series({}))                               ## Required 'foo' missing!
Traceback (most recent call last):
...
jsonschema.exceptions.ValidationError: 'foo' is a required property

Failed validating 'required' in schema:
    {'additionalProperties': False,
     'properties': {'foo': {}},
     'required': ['foo'],
     'type': 'object'}

On instance:
    Series([], dtype: float64)

class wltp.pandel.PathMaps[source]¶

Bases: builtins.object

Cascade prefix-mapping of json-paths to any values (here ModelOperations.

wltp.pandel.jsonpointer_parts(jsonpointer)[source]¶

Iterates over the jsonpointer parts.

Parameters:	jsonpointer (str) – a jsonpointer to resolve within document
Returns:	a generator over the parts of the json-pointer
Author:	Julian Berman, ankostis

wltp.pandel.resolve_jsonpointer(doc, jsonpointer, default=<object object at 0xb70f3fd8>)[source]¶

Resolve a jsonpointer within the referenced doc.

Parameters:	doc – the referrant document jsonpointer (str) – a jsonpointer to resolve within document
Returns:	the resolved doc-item or raises RefResolutionError
Author:	Julian Berman, ankostis

wltp.pandel.set_jsonpointer(doc, jsonpointer, value, object_factory=<class 'dict'>)[source]¶

Resolve a jsonpointer within the referenced doc.

Parameters:	doc – the referrant document jsonpointer (str) – a jsonpointer to the node to modify
Raises:	JsonPointerException (if jsonpointer empty, missing, invalid-contet)

6.4. Module: wltp.test.samples_db_tests¶

Compares the results of synthetic vehicles from JRC against pre-phase-1b Heinz’s tool.

• Run as Test-case to generate results for sample-vehicles.
• Run it as cmd-line to compare with Heinz’s results.

class wltp.test.samples_db_tests.ExperimentSampleVehs(methodName='runTest')[source]¶

Bases: unittest.case.TestCase

Compares a batch of vehicles with results obtained from “Official” implementation.

test1_AvgRPMs()[source]¶

Check mean-engine-speed diff with Heinz within some percent.

Results:

                   mean         std          min          max
python      1876.555626  146.755857  1652.457262  2220.657166
heinz       1892.048584  148.248303  1660.710716  2223.772904
diff_prcnt     0.008256    0.010170     0.004995     0.001403

test1_PMRatio()[source]¶

Check mean-engine-speed diff with Heinz within some percent for all PMRs.

Results:

                    gened_mean_rpm  heinz_mean_rpm  diff_prcnt  count
pmr
(40.759, 49.936]       1814.752308     1822.011660    0.004000      4
(49.936, 59.00401]     1861.137208     1879.822876    0.010040      4
(59.00401, 68.072]     2015.693195     2031.240237    0.007713      3
(68.072, 77.14]        1848.735584     1859.116047    0.005615      5
(77.14, 86.208]                NaN             NaN         NaN      0
(86.208, 95.276]       1786.879366     1807.764020    0.011688      5
(95.276, 104.344]      1956.288657     1980.523043    0.012388      3
(104.344, 113.412]     1929.718933     1947.787155    0.009363      3
(113.412, 122.48]      2033.321183     2051.602998    0.008991      1
(122.48, 131.548]      1781.487338     1781.591893    0.000059      1
(131.548, 140.616]             NaN             NaN         NaN      0
(140.616, 149.684]     1895.125082     1907.872848    0.006727      1

wltp.test.samples_db_tests.driver_weight = 70¶

For calculating unladen_mass.

6.5. Module: wltp.test.wltp_db_tests¶

Compares the results of a batch of wltp_db vehicles against phase-1b-alpha Heinz’s tool.

• Run as Test-case to generate results for sample-vehicles.
• Run it as cmd-line to compare with Heinz’s results.

class wltp.test.wltp_db_tests.WltpDbTests(methodName='runTest')[source]¶

Bases: unittest.case.TestCase

Compares a batch of vehicles with results obtained from “official” implementation.

test1_Downscale()[source]¶

Check mean-downscaled-velocity diff with Heinz within some percent.

### Comparison history ###

Force class3b, Phase-1b-beta(ver <= 0.0.8, Aug-2014) with Heinz maxt gear-time=2sec:

           python       heinz    diff_prcnt
count  378.000000  378.000000  0.000000e+00
mean    45.973545   46.189082  4.688300e-01
std      1.642335    1.126555 -4.578377e+01
min     35.866421   36.659117  2.210133e+00
25%     46.506718   46.504909 -3.892020e-03
50%     46.506718   46.506504 -4.620879e-04
75%     46.506718   46.506719  4.116024e-08
max     46.506718   46.506719  4.116024e-08

Not forcing class3b, honoring declared v_max & unladen_mass:

           python       heinz    diff_prcnt
count  382.000000  382.000000  0.000000e+00
mean    44.821337   44.846671  5.652189e-02
std      5.054214    5.050208 -7.933394e-02
min     28.091672   28.388418  1.056347e+00
25%     46.506718   46.504868 -3.978244e-03
50%     46.506718   46.506478 -5.162230e-04
75%     46.506718   46.506719  4.116033e-08
max     46.506718   46.506719  4.116033e-08

test2a_gear_diffs()[source]¶

Check diff-gears with Heinz stays within some percent.

### Comparison history ###

Class3b-Vehicles, Phase-1b-beta(ver <= 0.0.8, Aug-2014) with Heinz maxt gear-time=2sec:

             count       MEAN        STD  min  max
gears        23387  75.931818  56.921729    6  279
accell       19146  62.162338  48.831155    4  238
senza rules  16133  52.379870  35.858415   11  170

Separated test/unladen masses:

         diff_gears    diff_accel     diff_orig
count    378.000000    378.000000    378.000000
mean     104.965608     86.171958     90.235450
std      100.439783     82.613475    109.283901
min        6.000000      4.000000     11.000000
25%       36.250000     25.250000     23.000000
50%       69.000000     57.500000     51.000000
75%      142.000000    119.750000    104.750000
max      524.000000    404.000000    600.000000
sum    39677.000000  32573.000000  34109.000000
mean%      5.831423      4.787331      5.013081

Not forcing class3b, honoring declared v_max & unladen_mass:

         diff_gears    diff_accel     diff_orig
count    382.000000    382.000000    382.000000
mean      75.994764     63.633508     54.083770
std       58.290971     51.885162     38.762326
min        2.000000      2.000000      6.000000
25%       29.000000     22.000000     19.000000
50%       57.000000     48.500000     45.000000
75%      111.000000     97.000000     78.750000
max      279.000000    243.000000    173.000000
sum    29030.000000  24308.000000  20660.000000
mean%      4.221931      3.535195      3.004654

test2b_gear_diffs_transplanted()[source]¶

Check driveability-only diff-gears with Heinz stays within some percent.

### Comparison history ###

Force class3b, Phase-1b-beta(ver <= 0.0.8, Aug-2014) with Heinz maxt gear-time=2sec:

        diff_gears   diff_accel  diff_orig
count   378.000000   378.000000        378
mean     15.566138     5.634921          0
std      16.554295     8.136700          0
min       0.000000     0.000000          0
25%       5.000000     1.000000          0
50%      11.000000     3.000000          0
75%      19.750000     7.000000          0
max     123.000000    78.000000          0
sum    5884.000000  2130.000000          0
mean%     0.864785     0.313051          0

Not forcing class3b, honoring declared v_max & unladen_mass:

        diff_gears   diff_accel  diff_orig
count   382.000000   382.000000        382
mean     12.599476     4.651832          0
std      15.375930     7.566103          0
min       0.000000     0.000000          0
25%       4.000000     0.000000          0
50%       9.000000     2.000000          0
75%      15.000000     6.000000          0
max     123.000000    78.000000          0
sum    4813.000000  1777.000000          0
mean%     0.699971     0.258435          0

test3a_n_mean()[source]¶

Check mean-rpm diff with Heinz stays within some percent.

### Comparison history ###

Class3b-Vehicles, Phase-1b-beta(ver <= 0.0.8, Aug-2014) with Heinz maxt gear-time=2sec:

                   mean         std          min          max
python      1766.707825  410.762478  1135.458463  3217.428423
heinz       1759.851498  397.343498  1185.905053  3171.826208
diff_prcnt    -0.3896     -3.3772       4.4428      -1.4377

Separated test/unladen masses:

            python        heinz  diff_prcnt
count   378.000000   378.000000    0.000000
mean   1923.908119  1899.366431   -1.292099
std     628.998854   593.126296   -6.048047
min    1135.458463  1185.905053    4.442839
25%    1497.544940  1495.699889   -0.123357
50%    1740.927971  1752.668517    0.674384
75%    2121.459309  2111.876041   -0.453780
max    4965.206982  4897.154914   -1.389625

Not forcing class3b, honoring declared v_max & unladen_mass:

            python        heinz  diff_prcnt
count   382.000000   382.000000    0.000000
mean   1835.393402  1827.572965   -0.427914
std     476.687485   464.264779   -2.675781
min    1135.458463  1185.905053    4.442839
25%    1486.886555  1482.789006   -0.276341
50%    1731.983662  1739.781233    0.450210
75%    2024.534101  2018.716963   -0.288160
max    3741.849187  3750.927263    0.242609

test3b_n_mean_transplanted()[source]¶

Check driveability-only mean-rpm diff with Heinz stays within some percent.

### Comparison history ###

Force class3b, Phase-1b-beta(ver <= 0.0.8, Aug-2014) with Heinz maxt gear-time=2sec:

            python        heinz  diff_prcnt
count   378.000000   378.000000    0.000000
mean   1880.045112  1899.366431    1.027705
std     572.842493   593.126296    3.540904
min    1150.940393  1185.905053    3.037921
25%    1477.913404  1495.699889    1.203486
50%    1739.882957  1752.668517    0.734852
75%    2073.715015  2111.876041    1.840225
max    4647.136063  4897.154914    5.380063

Not forcing class3b, honoring declared v_max & unladen_mass:

            python        heinz  diff_prcnt
count   382.000000   382.000000    0.000000
mean   1818.519842  1827.572965    0.497829
std     469.276397   464.264779   -1.079474
min    1150.940393  1185.905053    3.037921
25%    1467.153958  1482.789006    1.065672
50%    1730.051632  1739.781233    0.562388
75%    2010.264758  2018.716963    0.420452
max    3704.999890  3750.927263    1.239605

test4a_n_mean__PMR()[source]¶

Check mean-rpm diff with Heinz stays within some percent for all PMRs.

### Comparison history ###

Force class3b, Phase-1b-beta(ver <= 0.0.8, Aug-2014) with Heinz maxt gear-time=2sec:

                    gened_mean_rpm  heinz_mean_rpm  diff_ratio  count
pmr
(9.973, 24.823]        1566.018469     1568.360963    0.001496     32
(24.823, 39.496]       1701.176128     1702.739797    0.000919     32
(39.496, 54.17]        1731.541637     1724.959671   -0.003816    106
(54.17, 68.843]        1894.477475     1877.786294   -0.008889     61
(68.843, 83.517]       1828.518522     1818.720627   -0.005387     40
(83.517, 98.191]       1824.060716     1830.482140    0.003520      3
(98.191, 112.864]      1794.673461     1792.693611   -0.001104     31
(112.864, 127.538]     3217.428423     3171.826208   -0.014377      1
(127.538, 142.211]     1627.952896     1597.571904   -0.019017      1
(142.211, 156.885]             NaN             NaN         NaN      0
(156.885, 171.558]             NaN             NaN         NaN      0
(171.558, 186.232]     1396.061758     1385.176569   -0.007858      1

Separated test/unladen masses:

                     gened_mean_rpm  heinz_mean_rpm  diff_prcnt  count
pmr
(11.504, 26.225]        1579.612698     1585.721306    0.386716     28
(26.225, 40.771]        1706.865069     1700.689983   -0.363093     41
(40.771, 55.317]        1866.150857     1841.779091   -1.323273    119
(55.317, 69.863]        2122.662626     2085.262950   -1.793523    122
(69.863, 84.409]        2228.282795     2171.952804   -2.593518     29
(84.409, 98.955]        1783.316413     1787.378401    0.227777      4
(98.955, 113.501]       1718.157828     1718.516147    0.020855     31
(113.501, 128.0475]     2005.415058     1954.763742   -2.591173      2
(128.0475, 142.594]     1566.601860     1553.383676   -0.850928      1
(142.594, 157.14]               NaN             NaN         NaN      0
(157.14, 171.686]               NaN             NaN         NaN      0
(171.686, 186.232]      1396.061758     1385.176569   -0.785834      1

Not forcing class3b, honoring declared v_max & unladen_mass:

                    gened_mean_rpm  heinz_mean_rpm  diff_prcnt  count
pmr
(9.973, 24.823]        1560.010258     1563.836656    0.245280     33
(24.823, 39.496]       1725.209986     1725.004638   -0.011904     34
(39.496, 54.17]        1737.811065     1730.770088   -0.406812    123
(54.17, 68.843]        1996.999520     1983.753219   -0.667739     94
(68.843, 83.517]       2051.088434     2034.594136   -0.810692     59
(83.517, 98.191]       1964.832555     1958.081066   -0.344801      4
(98.191, 112.864]      1682.122484     1684.443875    0.138004     31
(112.864, 127.538]     2718.877009     2687.055802   -1.184241      2
(127.538, 142.211]     1660.925042     1668.155469    0.435325      1
(142.211, 156.885]             NaN             NaN         NaN      0
(156.885, 171.558]             NaN             NaN         NaN      0
(171.558, 186.232]     1396.061758     1385.176569   -0.785834      1
Mean: 0.419219429398

test4b_n_mean__PMR_transplanted()[source]¶

Check driveability-only mean-rpm diff with Heinz stays within some percent for all PMRs.

### Comparison history ###

Force class3b, Phase-1b-beta(ver <= 0.0.8, Aug-2014) with Heinz maxt gear-time=2sec:

                    gened_mean_rpm  heinz_mean_rpm  diff_prcnt  count
pmr
(9.973, 24.823]        1557.225037     1568.360963    0.715113     32
(24.823, 39.496]       1686.859826     1696.482640    0.570457     34
(39.496, 54.17]        1771.670097     1789.409819    1.001299    120
(54.17, 68.843]        2133.400050     2165.214662    1.491263     94
(68.843, 83.517]       2020.903728     2043.741660    1.130085     59
(83.517, 98.191]       1886.836446     1890.040533    0.169813      4
(98.191, 112.864]      1788.434592     1792.693611    0.238142     31
(112.864, 127.538]     2580.884314     2568.011660   -0.501269      2
(127.538, 142.211]     1581.625191     1597.571904    1.008249      1
(142.211, 156.885]             NaN             NaN         NaN      0
(156.885, 171.558]             NaN             NaN         NaN      0
(171.558, 186.232]     1367.068837     1385.176569    1.324566      1

Separated test/unladen masses:

                     gened_mean_rpm  heinz_mean_rpm  diff_prcnt  count
pmr
(11.504, 26.225]        1572.733597     1585.721306    0.825805     28
(26.225, 40.771]        1690.081663     1700.689983    0.627681     41
(40.771, 55.317]        1821.319706     1841.779091    1.123327    119
(55.317, 69.863]        2060.507029     2085.262950    1.201448    122
(69.863, 84.409]        2142.964427     2171.952804    1.352723     29
(84.409, 98.955]        1783.214173     1787.378401    0.233524      4
(98.955, 113.501]       1713.473617     1718.516147    0.294287     31
(113.501, 128.0475]     1950.373771     1954.763742    0.225084      2
(128.0475, 142.594]     1543.937285     1553.383676    0.611838      1
(142.594, 157.14]               NaN             NaN         NaN      0
(157.14, 171.686]               NaN             NaN         NaN      0
(171.686, 186.232]      1367.068837     1385.176569    1.324566      1

Not forcing class3b, honoring declared v_max & unladen_mass:

                    gened_mean_rpm  heinz_mean_rpm  diff_prcnt  count
pmr
(9.973, 24.823]        1551.901645     1563.836656    0.769057     33
(24.823, 39.496]       1713.382835     1725.004638    0.678296     34
(39.496, 54.17]        1722.174466     1730.770088    0.499114    123
(54.17, 68.843]        1974.768859     1983.753219    0.454958     94
(68.843, 83.517]       2026.630271     2034.594136    0.392961     59
(83.517, 98.191]       1954.817179     1958.081066    0.166966      4
(98.191, 112.864]      1676.678357     1684.443875    0.463149     31
(112.864, 127.538]     2678.973439     2687.055802    0.301696      2
(127.538, 142.211]     1658.577318     1668.155469    0.577492      1
(142.211, 156.885]             NaN             NaN         NaN      0
(156.885, 171.558]             NaN             NaN         NaN      0
(171.558, 186.232]     1367.068837     1385.176569    1.324566      1
Mean diff_prcnt: 0.469021296461

test5a_n_mean__gear()[source]¶

Check mean-rpm diff% with Heinz stays within some percent for all gears.

### Comparison history ###

Force class3b, Phase-1b-beta(ver <= 0.0.8, Aug-2014) with Heinz maxt gear-time=2sec:

n_mean      python        heinz      diff%
gear
0       732.358286   804.656085  -9.925769
1       870.080494  1177.547512 -44.450903
2      1789.787609  1650.383967   6.520319
3      1921.271483  1761.172027   7.804359
4      1990.286402  1886.563262   5.401895
5      2138.445024  2112.552162   1.892950
6      2030.970322  1987.865039   2.228276

Not forcing class3b, honoring declared v_max & unladen_mass:

gear
0        735.143823   808.795812 -10.052865
1        799.834530  1139.979330 -47.027383
2       1598.773915  1582.431975   1.119054
3       1793.617644  1691.589756   5.768020
4       1883.863510  1796.957457   5.024360
5       2095.211754  2052.059948   2.430360
6       2033.663975  1990.344346   2.238421

test5b_n_mean__gear_transplanted()[source]¶

Check mean-rpm diff% with Heinz stays within some percent for all gears.

### Comparison history ###

Force class3b, Phase-1b-beta(ver <= 0.0.8, Aug-2014) with Heinz maxt gear-time=2sec:

n_mean      python        heinz      diff%
gear
0       732.357001   804.656085  -9.926855
1       966.022039  1177.547512 -24.409425
2      1678.578373  1650.383967   1.616768
3      1791.644768  1761.172027   1.700642
4      1883.504933  1886.563262   0.119165
5      2099.218160  2112.552162  -0.320293
6      1985.732086  1987.865039  -0.096754

Not forcing class3b, honoring declared v_max & unladen_mass:

n_mean       python        heinz      diff%
gear
0        735.077116   808.795812 -10.065886
1        932.586982  1139.979330 -24.285307
2       1606.040896  1582.431975   1.379144
3       1721.141364  1691.589756   1.686708
4       1803.212699  1796.957457   0.370703
5       2053.822313  2052.059948   0.142138
6       1988.195381  1990.344346  -0.097482

wltp.test.wltp_db_tests._file_pairs(fname_glob)[source]¶

Generates pairs of files to compare, skipping non-existent and those with mismatching #_of_rows.

Example:

>>> for (veh_num, df_g, df_h) in _file_pairs('wltp_db_vehicles-00*.csv')
        pass

wltp.test.wltp_db_tests.aggregate_single_columns_means(gened_column, heinz_column)[source]¶

Runs experiments and aggregates mean-values from one column of each (gened, heinz) file-sets.

wltp.test.wltp_db_tests.driver_weight = 70¶

For calculating unladen_mass.

wltp.test.wltp_db_tests.vehicles_applicator(fname_glob, pair_func)[source]¶

Applies the fun onto a pair of (generated, heinz) files for each tested-vehicle in the glob and appends results to list, preffixed by veh_num.

Parameters:	pair_func – signature: func(veh_no, gened_df, heinz_df)–>sequence_of_numbers
Returns:	a dataframe with the columns returned from the pair_func, row_indexed by veh_num