Skip to content

Result Augmentation

SRToolkit.evaluation.result_augmentation

Concrete ResultAugmenter implementations that post-process SR results with additional information.

Available augmenters: ExpressionToLatex, ExpressionSimplifier, RMSE, BED, R2. Custom augmenters work automatically without registration — just subclass ResultAugmenter and implement the required methods.

ExpressionToLatex 

ExpressionToLatex(symbol_library: SymbolLibrary, scope: str = 'top', verbose: bool = False, name: str = 'ExpressionToLatex')

Bases: ResultAugmenter

Converts expressions inside the results to LaTeX strings.

Parameters:

Name Type Description Default
symbol_library SymbolLibrary

Symbol library used to produce LaTeX templates for each token.

 required
scope str

Which expressions to convert.

  • "best": only the best expression.
  • "top": the best expression and all top-k models.
  • "all": everything in "top" plus all evaluated expressions.
 'top'
verbose bool

If True, emits a warning when LaTeX conversion fails for an expression. Default False.

 False
name str

Key used in augmentations dict of EvalResult and ModelResult. Default "ExpressionToLatex".

 'ExpressionToLatex'
Source code in SRToolkit/evaluation/result_augmentation.py 
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
def __init__(
    self,
    symbol_library: SymbolLibrary,
    scope: str = "top",
    verbose: bool = False,
    name: str = "ExpressionToLatex",
) -> None:
    """
    Converts expressions inside the results to LaTeX strings.

    Args:
        symbol_library: Symbol library used to produce LaTeX templates for each token.
        scope: Which expressions to convert.

            - ``"best"``: only the best expression.
            - ``"top"``: the best expression and all top-k models.
            - ``"all"``: everything in ``"top"`` plus all evaluated expressions.
        verbose: If ``True``, emits a warning when LaTeX conversion fails for an expression.
            Default ``False``.
        name: Key used in
            ``augmentations`` dict of [EvalResult][SRToolkit.utils.types.EvalResult] and
            [ModelResult][SRToolkit.utils.types.ModelResult].
            Default ``"ExpressionToLatex"``.
    """
    super().__init__(name)
    self.symbol_library = symbol_library

    if scope not in ["best", "top", "all"]:
        raise Exception(f"[RMSE augmenter] Invalid scope: {scope}. Must be one of 'best', 'top', 'all'.")
    self.scope = scope

    self.verbose = verbose

write_results 

write_results(results: EvalResult) -> None

Write LaTeX representations into results and its models.

Stores {"best_expr_latex": ...} in EvalResult augmentations. Also stores {"expr_latex": ...} in each model's augmentations when scope is "top" or "all".

Parameters:

Name Type Description Default
results EvalResult

The EvalResult to augment.

 required
Source code in SRToolkit/evaluation/result_augmentation.py 
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
def write_results(self, results: EvalResult) -> None:
    """
    Write LaTeX representations into *results* and its models.

    Stores ``{"best_expr_latex": ...}`` in
    [EvalResult][SRToolkit.utils.types.EvalResult] ``augmentations``.
    Also stores ``{"expr_latex": ...}`` in each model's augmentations when
    ``scope`` is ``"top"`` or ``"all"``.

    Args:
        results: The [EvalResult][SRToolkit.utils.types.EvalResult] to augment.
    """
    eval_data: Dict[str, Any] = {}
    try:
        eval_data["best_expr_latex"] = tokens_to_tree(results.top_models[0].expr, self.symbol_library).to_latex(
            self.symbol_library
        )
    except Exception as e:
        if self.verbose:
            warnings.warn(f"Unable to convert best expression to LaTeX: {e}")
    results.add_augmentation(self.name, eval_data, self._type)

    if self.scope == "top" or self.scope == "all":
        for model in results.top_models:
            try:
                model.add_augmentation(
                    self.name,
                    {"expr_latex": tokens_to_tree(model.expr, self.symbol_library).to_latex(self.symbol_library)},
                    self._type,
                )
            except Exception as e:
                if self.verbose:
                    warnings.warn(f"Unable to convert expression {''.join(model.expr)} to LaTeX: {e}")

    if self.scope == "all":
        for model in results.all_models:
            try:
                model.add_augmentation(
                    self.name,
                    {"expr_latex": tokens_to_tree(model.expr, self.symbol_library).to_latex(self.symbol_library)},
                    self._type,
                )
            except Exception as e:
                if self.verbose:
                    warnings.warn(f"Unable to convert expression {''.join(model.expr)} to LaTeX: {e}")

format_eval_result classmethod 

format_eval_result(data: Dict[str, Any]) -> str

Format experiment-level LaTeX augmentation data for display.

Parameters:

Name Type Description Default
data Dict[str, Any]

Augmentation dict containing "best_expr_latex".

 required

Returns:

Type Description
str

A human-readable string, or empty string if no data is present.
Source code in SRToolkit/evaluation/result_augmentation.py 
104
105
106
107
108
109
110
111
112
113
114
115
116
@classmethod
def format_eval_result(cls, data: Dict[str, Any]) -> str:
    """
    Format experiment-level LaTeX augmentation data for display.

    Args:
        data: Augmentation dict containing ``"best_expr_latex"``.

    Returns:
        A human-readable string, or empty string if no data is present.
    """
    latex = data.get("best_expr_latex", "")
    return f"LaTeX of the best expression: {latex}" if latex else ""

format_model_result classmethod 

format_model_result(data: Dict[str, Any]) -> str

Format per-model LaTeX augmentation data for display.

Parameters:

Name Type Description Default
data Dict[str, Any]

Augmentation dict containing "expr_latex".

 required

Returns:

Type Description
str

A human-readable string, or empty string if no data is present.
Source code in SRToolkit/evaluation/result_augmentation.py 
118
119
120
121
122
123
124
125
126
127
128
129
130
@classmethod
def format_model_result(cls, data: Dict[str, Any]) -> str:
    """
    Format per-model LaTeX augmentation data for display.

    Args:
        data: Augmentation dict containing ``"expr_latex"``.

    Returns:
        A human-readable string, or empty string if no data is present.
    """
    latex = data.get("expr_latex", "")
    return f"LaTeX: {latex}" if latex else ""

to_dict 

to_dict(base_path: str, name: str) -> dict

Creates a dictionary representation of the ExpressionToLatex augmenter.

Parameters:

Name Type Description Default
base_path str

Unused and ignored

 required
name str

Unused and ignored

 required

Returns:

Type Description
dict

A dictionary containing the necessary information to recreate the augmenter.
Source code in SRToolkit/evaluation/result_augmentation.py 
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
def to_dict(self, base_path: str, name: str) -> dict:
    """
    Creates a dictionary representation of the ExpressionToLatex augmenter.

    Args:
        base_path: Unused and ignored
        name: Unused and ignored

    Returns:
        A dictionary containing the necessary information to recreate the augmenter.
    """
    return {
        "format_version": 1,
        "type": "ExpressionToLatex",
        "name": self.name,
        "symbol_library": self.symbol_library.to_dict(),
        "scope": self.scope,
        "verbose": self.verbose,
    }

from_dict staticmethod 

from_dict(data: dict) -> ExpressionToLatex

Creates an instance of the ExpressionToLatex augmenter from a dictionary.

Parameters:

Name Type Description Default
data dict

A dictionary containing the necessary information to recreate the augmenter.

 required

Returns:

Type Description
ExpressionToLatex

An instance of the ExpressionToLatex augmenter.
Source code in SRToolkit/evaluation/result_augmentation.py 
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
@staticmethod
def from_dict(data: dict) -> "ExpressionToLatex":
    """
    Creates an instance of the ExpressionToLatex augmenter from a dictionary.

    Args:
        data: A dictionary containing the necessary information to recreate the augmenter.

    Returns:
        An instance of the ExpressionToLatex augmenter.
    """
    if data.get("format_version", 1) != 1:
        raise ValueError(
            f"[ExpressionToLatex.from_dict] Unsupported format_version: {data.get('format_version')!r}. Expected 1."
        )
    return ExpressionToLatex(
        symbol_library=data["symbol_library"],
        scope=data["scope"],
        verbose=data["verbose"],
        name=data["name"],
    )

ExpressionSimplifier 

ExpressionSimplifier(symbol_library: SymbolLibrary, scope: str = 'top', verbose: bool = False, name: str = 'ExpressionSimplifier')

Bases: ResultAugmenter

Algebraically simplifies expressions inside the results using SymPy.

Parameters:

Name Type Description Default
symbol_library SymbolLibrary

Symbol library used by the simplifier to resolve token types.

 required
scope str

Which expressions to simplify.

  • "best": only the best expression.
  • "top": the best expression and all top-k models.
  • "all": everything in "top" plus all evaluated expressions.
 'top'
verbose bool

If True, emits a warning when simplification fails for an expression. Default False.

 False
name str

Key used in augmentations dict of EvalResult and ModelResult. Default "ExpressionSimplifier".

 'ExpressionSimplifier'
Source code in SRToolkit/evaluation/result_augmentation.py 
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
def __init__(
    self,
    symbol_library: SymbolLibrary,
    scope: str = "top",
    verbose: bool = False,
    name: str = "ExpressionSimplifier",
) -> None:
    """
    Algebraically simplifies expressions inside the results using SymPy.

    Args:
        symbol_library: Symbol library used by the simplifier to resolve token types.
        scope: Which expressions to simplify.

            - ``"best"``: only the best expression.
            - ``"top"``: the best expression and all top-k models.
            - ``"all"``: everything in ``"top"`` plus all evaluated expressions.
        verbose: If ``True``, emits a warning when simplification fails for an expression.
            Default ``False``.
        name: Key used in
            ``augmentations`` dict of [EvalResult][SRToolkit.utils.types.EvalResult] and
            [ModelResult][SRToolkit.utils.types.ModelResult].
            Default ``"ExpressionSimplifier"``.
    """
    super().__init__(name)
    self.symbol_library = symbol_library

    if scope not in ["best", "top", "all"]:
        raise Exception(f"[RMSE augmenter] Invalid scope: {scope}. Must be one of 'best', 'top', 'all'.")
    self.scope = scope

    self.verbose = verbose

write_results 

write_results(results: EvalResult) -> None

Write simplified expressions into results and its models.

Stores {"simplified_best_expr": ...} in EvalResult augmentations if simplification succeeds. Also stores {"simplified_expr": ...} in each model's augmentations when scope is "top" or "all".

Parameters:

Name Type Description Default
results EvalResult

The EvalResult to augment.

 required
Source code in SRToolkit/evaluation/result_augmentation.py 
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
def write_results(self, results: EvalResult) -> None:
    """
    Write simplified expressions into *results* and its models.

    Stores ``{"simplified_best_expr": ...}`` in
    [EvalResult][SRToolkit.utils.types.EvalResult] ``augmentations`` if
    simplification succeeds. Also stores ``{"simplified_expr": ...}`` in each model's
    augmentations when ``scope`` is ``"top"`` or ``"all"``.

    Args:
        results: The [EvalResult][SRToolkit.utils.types.EvalResult] to augment.
    """
    eval_data: Dict[str, Any] = {}
    try:
        simplified_expr = simplify(results.top_models[0].expr, self.symbol_library)
        if isinstance(simplified_expr, list):
            eval_data["simplified_best_expr"] = "".join(simplified_expr)
        elif isinstance(simplified_expr, Node):
            eval_data["simplified_best_expr"] = "".join(simplified_expr.to_list(self.symbol_library))
        else:
            raise Exception(f"Simplified expression is not a list or Node: {simplified_expr}")
    except Exception as e:
        if self.verbose:
            warnings.warn(f"Unable to simplify {results.best_expr}: {e}")
    results.add_augmentation(self.name, eval_data, self._type)

    if self.scope == "top" or self.scope == "all":
        for model in results.top_models:
            top_model_data: Dict[str, Any] = {}
            try:
                simplified_expr = simplify(model.expr, self.symbol_library)
                if isinstance(simplified_expr, list):
                    top_model_data["simplified_expr"] = "".join(simplified_expr)
                elif isinstance(simplified_expr, Node):
                    top_model_data["simplified_expr"] = "".join(simplified_expr.to_list(self.symbol_library))
                else:
                    raise Exception(f"Simplified expression is not a list or Node: {simplified_expr}")
            except Exception as e:
                if self.verbose:
                    warnings.warn(f"Unable to simplify {''.join(model.expr)}: {e}")
            model.add_augmentation(self.name, top_model_data, self._type)

    if self.scope == "all":
        for model in results.all_models:
            all_model_data: Dict[str, Any] = {}
            try:
                simplified_expr = simplify(model.expr, self.symbol_library)
                if isinstance(simplified_expr, list):
                    all_model_data["simplified_expr"] = "".join(simplified_expr)
                elif isinstance(simplified_expr, Node):
                    all_model_data["simplified_expr"] = "".join(simplified_expr.to_list(self.symbol_library))
                else:
                    raise Exception(f"Simplified expression is not a list or Node: {simplified_expr}")
            except Exception as e:
                if self.verbose:
                    warnings.warn(f"Unable to simplify {''.join(model.expr)}: {e}")
            model.add_augmentation(self.name, all_model_data, self._type)

format_eval_result classmethod 

format_eval_result(data: Dict[str, Any]) -> str

Format experiment-level simplification data for display.

Parameters:

Name Type Description Default
data Dict[str, Any]

Augmentation dict containing "simplified_best_expr".

 required

Returns:

Type Description
str

A human-readable string, or empty string if no data is present.
Source code in SRToolkit/evaluation/result_augmentation.py 
267
268
269
270
271
272
273
274
275
276
277
278
279
@classmethod
def format_eval_result(cls, data: Dict[str, Any]) -> str:
    """
    Format experiment-level simplification data for display.

    Args:
        data: Augmentation dict containing ``"simplified_best_expr"``.

    Returns:
        A human-readable string, or empty string if no data is present.
    """
    simplified = data.get("simplified_best_expr", "")
    return f"Simplified: {simplified}" if simplified else ""

format_model_result classmethod 

format_model_result(data: Dict[str, Any]) -> str

Format per-model simplification data for display.

Parameters:

Name Type Description Default
data Dict[str, Any]

Augmentation dict containing "simplified_expr".

 required

Returns:

Type Description
str

A human-readable string, or empty string if no data is present.
Source code in SRToolkit/evaluation/result_augmentation.py 
281
282
283
284
285
286
287
288
289
290
291
292
293
@classmethod
def format_model_result(cls, data: Dict[str, Any]) -> str:
    """
    Format per-model simplification data for display.

    Args:
        data: Augmentation dict containing ``"simplified_expr"``.

    Returns:
        A human-readable string, or empty string if no data is present.
    """
    simplified = data.get("simplified_expr", "")
    return f"Simplified: {simplified}" if simplified else ""

to_dict 

to_dict(base_path: str, name: str) -> dict

Creates a dictionary representation of the ExpressionSimplifier augmenter.

Parameters:

Name Type Description Default
base_path str

Unused and ignored

 required
name str

Unused and ignored

 required

Returns:

Type Description
dict

A dictionary containing the necessary information to recreate the augmenter.
Source code in SRToolkit/evaluation/result_augmentation.py 
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
def to_dict(self, base_path: str, name: str) -> dict:
    """
    Creates a dictionary representation of the ExpressionSimplifier augmenter.

    Args:
        base_path: Unused and ignored
        name: Unused and ignored

    Returns:
        A dictionary containing the necessary information to recreate the augmenter.
    """
    return {
        "format_version": 1,
        "type": "ExpressionSimplifier",
        "name": self.name,
        "symbol_library": self.symbol_library.to_dict(),
        "scope": self.scope,
        "verbose": self.verbose,
    }

from_dict staticmethod 

from_dict(data: dict) -> ExpressionSimplifier

Creates an instance of the ExpressionSimplifier augmenter from a dictionary.

Parameters:

Name Type Description Default
data dict

A dictionary containing the necessary information to recreate the augmenter.

 required

Returns: An instance of the ExpressionSimplifier augmenter.

Source code in SRToolkit/evaluation/result_augmentation.py 
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
@staticmethod
def from_dict(data: dict) -> "ExpressionSimplifier":
    """
    Creates an instance of the ExpressionSimplifier augmenter from a dictionary.

    Args:
        data: A dictionary containing the necessary information to recreate the augmenter.
    Returns:
        An instance of the ExpressionSimplifier augmenter.
    """
    if data.get("format_version", 1) != 1:
        raise ValueError(
            f"[ExpressionSimplifier.from_dict] Unsupported format_version: {data.get('format_version')!r}. Expected 1."
        )
    return ExpressionSimplifier(
        symbol_library=data["symbol_library"],
        scope=data["scope"],
        verbose=data["verbose"],
        name=data["name"],
    )

RMSE 

RMSE(evaluator: SR_evaluator, scope: str = 'top', name: str = 'RMSE')

Bases: ResultAugmenter

Computes RMSE for the top models using a separate evaluator (e.g. a held-out test set).

Parameters:

Name Type Description Default
evaluator SR_evaluator

SR_evaluator used to score the models. Must be initialized with ranking_function="rmse" and a non-None y.

 required
scope str

Which expressions to score.

  • "best": only the best expression.
  • "top": the best expression and all top-k models.
  • "all": everything in "top" plus all evaluated expressions.
 'top'
name str

Key used in augmentations dict of EvalResult and ModelResult. Default "RMSE".

 'RMSE'

Raises:

Type Description
Exception

If evaluator.ranking_function != "rmse" or evaluator.y is None.
Source code in SRToolkit/evaluation/result_augmentation.py 
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
def __init__(self, evaluator: SR_evaluator, scope: str = "top", name: str = "RMSE") -> None:  # noqa: F821
    """
    Computes RMSE for the top models using a separate evaluator (e.g. a held-out test set).

    Args:
        evaluator: [SR_evaluator][SRToolkit.evaluation.sr_evaluator.SR_evaluator] used to
            score the models. Must be initialized with ``ranking_function="rmse"`` and a
            non-``None`` ``y``.
        scope: Which expressions to score.

            - ``"best"``: only the best expression.
            - ``"top"``: the best expression and all top-k models.
            - ``"all"``: everything in ``"top"`` plus all evaluated expressions.
        name: Key used in
            ``augmentations`` dict of [EvalResult][SRToolkit.utils.types.EvalResult] and
            [ModelResult][SRToolkit.utils.types.ModelResult].
            Default ``"RMSE"``.

    Raises:
        Exception: If ``evaluator.ranking_function != "rmse"`` or ``evaluator.y is None``.
    """
    super().__init__(name)
    self.evaluator = evaluator

    if scope not in ["best", "top", "all"]:
        raise Exception(f"[RMSE augmenter] Invalid scope: {scope}. Must be one of 'best', 'top', 'all'.")
    self.scope = scope

    if self.evaluator.ranking_function != "rmse":
        raise Exception("[RMSE augmenter] Ranking function of the evaluator must be set to 'rmse' to compute RMSE.")
    if self.evaluator.y is None:
        raise Exception("[RMSE augmenter] y in the evaluator must not be None to compute RMSE.")

write_results 

write_results(results: EvalResult) -> None

Write RMSE scores into results and its models.

Stores {"min_error": ...} in EvalResult augmentations and {"error": ..., "parameters": ...} in each model's augmentations when scope is "top" or "all".

Parameters:

Name Type Description Default
results EvalResult

The EvalResult to augment.

 required
Source code in SRToolkit/evaluation/result_augmentation.py 
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
def write_results(self, results: EvalResult) -> None:
    """
    Write RMSE scores into *results* and its models.

    Stores ``{"min_error": ...}`` in
    [EvalResult][SRToolkit.utils.types.EvalResult] ``augmentations`` and
    ``{"error": ..., "parameters": ...}`` in each model's augmentations when ``scope``
    is ``"top"`` or ``"all"``.

    Args:
        results: The [EvalResult][SRToolkit.utils.types.EvalResult] to augment.
    """
    eval_data: Dict[str, Any] = {"min_error": self.evaluator.evaluate_expr(results.top_models[0].expr)}
    results.add_augmentation(self.name, eval_data, self._type)

    if self.scope == "top" or self.scope == "all":
        for model in results.top_models:
            key = "".join(model.expr)
            top_model_data: Dict[str, Any] = {
                "error": self.evaluator.evaluate_expr(model.expr),
                "parameters": self.evaluator.models[key].parameters,
            }
            model.add_augmentation(self.name, top_model_data, self._type)

    if self.scope == "all":
        for model in results.all_models:
            key = "".join(model.expr)
            all_model_data: Dict[str, Any] = {
                "error": self.evaluator.evaluate_expr(model.expr),
                "parameters": self.evaluator.models[key].parameters,
            }
            model.add_augmentation(self.name, all_model_data, self._type)

format_eval_result classmethod 

format_eval_result(data: Dict[str, Any]) -> str

Format experiment-level RMSE data for display.

Parameters:

Name Type Description Default
data Dict[str, Any]

Augmentation dict containing "min_error".

 required

Returns:

Type Description
str

A human-readable string, or empty string if no data is present.
Source code in SRToolkit/evaluation/result_augmentation.py 
404
405
406
407
408
409
410
411
412
413
414
415
416
@classmethod
def format_eval_result(cls, data: Dict[str, Any]) -> str:
    """
    Format experiment-level RMSE data for display.

    Args:
        data: Augmentation dict containing ``"min_error"``.

    Returns:
        A human-readable string, or empty string if no data is present.
    """
    val = data.get("min_error", "")
    return f"Test RMSE: {val}" if val != "" else ""

format_model_result classmethod 

format_model_result(data: Dict[str, Any]) -> str

Format per-model RMSE data for display.

Parameters:

Name Type Description Default
data Dict[str, Any]

Augmentation dict containing "error" and optionally "parameters".

 required

Returns:

Type Description
str

A human-readable string with RMSE and fitted parameters.
Source code in SRToolkit/evaluation/result_augmentation.py 
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
@classmethod
def format_model_result(cls, data: Dict[str, Any]) -> str:
    """
    Format per-model RMSE data for display.

    Args:
        data: Augmentation dict containing ``"error"`` and optionally ``"parameters"``.

    Returns:
        A human-readable string with RMSE and fitted parameters.
    """
    parts = [f"RMSE={data['error']:.6g}"]
    if "parameters" in data and data["parameters"] is not None:
        parts.append(f"params={np.round(data['parameters'], 4).tolist()}")
    return ", ".join(parts)

to_dict 

to_dict(base_path: str, name: str) -> dict

Creates a dictionary representation of the RMSE augmenter.

Parameters:

Name Type Description Default
base_path str

Used to save the data of the evaluator to disk.

 required
name str

Used to save the data of the evaluator to disk.

 required

Returns:

Type Description
dict

A dictionary containing the necessary information to recreate the augmenter.
Source code in SRToolkit/evaluation/result_augmentation.py 
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
def to_dict(self, base_path: str, name: str) -> dict:
    """
    Creates a dictionary representation of the RMSE augmenter.

    Args:
        base_path: Used to save the data of the evaluator to disk.
        name: Used to save the data of the evaluator to disk.

    Returns:
        A dictionary containing the necessary information to recreate the augmenter.
    """
    return {
        "format_version": 1,
        "name": self.name,
        "type": "RMSE",
        "scope": self.scope,
        "evaluator": self.evaluator.to_dict(base_path, name + "_RMSE_augmenter"),
    }

from_dict staticmethod 

from_dict(data: dict) -> RMSE

Creates an instance of the RMSE augmenter from a dictionary.

Parameters:

Name Type Description Default
data dict

A dictionary containing the necessary information to recreate the augmenter.

 required

Returns:

Type Description
RMSE

An instance of the RMSE augmenter.
Source code in SRToolkit/evaluation/result_augmentation.py 
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
@staticmethod
def from_dict(data: dict) -> "RMSE":
    """
    Creates an instance of the RMSE augmenter from a dictionary.

    Args:
        data: A dictionary containing the necessary information to recreate the augmenter.

    Returns:
        An instance of the RMSE augmenter.
    """
    if data.get("format_version", 1) != 1:
        raise ValueError(
            f"[RMSE.from_dict] Unsupported format_version: {data.get('format_version')!r}. Expected 1."
        )
    evaluator = SR_evaluator.from_dict(data["evaluator"])
    return RMSE(evaluator, scope=data["scope"], name=data["name"])

BED 

BED(evaluator: SR_evaluator, scope: str = 'top', name: str = 'BED')

Bases: ResultAugmenter

Computes BED for the top models using a separate evaluator (e.g. a held-out test set).

Parameters:

Name Type Description Default
evaluator SR_evaluator

SR_evaluator used to score the models. Must be initialized with ranking_function="bed".

 required
scope str

Which expressions to score.

  • "best": only the best expression.
  • "top": the best expression and all top-k models.
  • "all": everything in "top" plus all evaluated expressions.
 'top'
name str

Key used in augmentations dict of EvalResult and ModelResult. Default "BED".

 'BED'

Raises:

Type Description
Exception

If evaluator.ranking_function != "bed".
Source code in SRToolkit/evaluation/result_augmentation.py 
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
def __init__(self, evaluator: SR_evaluator, scope: str = "top", name: str = "BED") -> None:  # noqa: F821
    """
    Computes BED for the top models using a separate evaluator (e.g. a held-out test set).

    Args:
        evaluator: [SR_evaluator][SRToolkit.evaluation.sr_evaluator.SR_evaluator] used to
            score the models. Must be initialized with ``ranking_function="bed"``.
        scope: Which expressions to score.

            - ``"best"``: only the best expression.
            - ``"top"``: the best expression and all top-k models.
            - ``"all"``: everything in ``"top"`` plus all evaluated expressions.
        name: Key used in
            ``augmentations`` dict of [EvalResult][SRToolkit.utils.types.EvalResult] and
            [ModelResult][SRToolkit.utils.types.ModelResult].
            Default ``"BED"``.

    Raises:
        Exception: If ``evaluator.ranking_function != "bed"``.
    """
    super().__init__(name)
    self.evaluator = evaluator

    if scope not in ["best", "top", "all"]:
        raise Exception(f"[BED augmenter] Invalid scope: {scope}. Must be one of 'best', 'top', 'all'.")
    self.scope = scope

    if self.evaluator.ranking_function != "bed":
        raise Exception("[BED augmenter] Ranking function of the evaluator must be set to 'bed' to compute BED.")

write_results 

write_results(results: EvalResult) -> None

Write BED scores into results and its models.

Stores {"best_expr_bed": ...} in EvalResult augmentations and {"bed": ...} in each model's augmentations when scope is "top" or "all".

Parameters:

Name Type Description Default
results EvalResult

The EvalResult to augment.

 required
Source code in SRToolkit/evaluation/result_augmentation.py 
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
def write_results(
    self,
    results: EvalResult,
) -> None:
    """
    Write BED scores into *results* and its models.

    Stores ``{"best_expr_bed": ...}`` in
    [EvalResult][SRToolkit.utils.types.EvalResult] ``augmentations`` and
    ``{"bed": ...}`` in each model's augmentations when ``scope`` is ``"top"`` or ``"all"``.

    Args:
        results: The [EvalResult][SRToolkit.utils.types.EvalResult] to augment.
    """
    eval_data: Dict[str, Any] = {"best_expr_bed": self.evaluator.evaluate_expr(results.top_models[0].expr)}
    results.add_augmentation(self.name, eval_data, self._type)

    if self.scope == "top" or self.scope == "all":
        for model in results.top_models:
            top_model_data: Dict[str, Any] = {"bed": self.evaluator.evaluate_expr(model.expr)}
            model.add_augmentation(self.name, top_model_data, self._type)

    if self.scope == "all":
        for model in results.all_models:
            all_model_data: Dict[str, Any] = {"bed": self.evaluator.evaluate_expr(model.expr)}
            model.add_augmentation(self.name, all_model_data, self._type)

format_eval_result classmethod 

format_eval_result(data: Dict[str, Any]) -> str

Format experiment-level BED data for display.

Parameters:

Name Type Description Default
data Dict[str, Any]

Augmentation dict containing "best_expr_bed".

 required

Returns:

Type Description
str

A human-readable string, or empty string if no data is present.
Source code in SRToolkit/evaluation/result_augmentation.py 
530
531
532
533
534
535
536
537
538
539
540
541
542
@classmethod
def format_eval_result(cls, data: Dict[str, Any]) -> str:
    """
    Format experiment-level BED data for display.

    Args:
        data: Augmentation dict containing ``"best_expr_bed"``.

    Returns:
        A human-readable string, or empty string if no data is present.
    """
    val = data.get("best_expr_bed", "")
    return f"Test BED: {val}" if val != "" else ""

format_model_result classmethod 

format_model_result(data: Dict[str, Any]) -> str

Format per-model BED data for display.

Parameters:

Name Type Description Default
data Dict[str, Any]

Augmentation dict containing "bed".

 required

Returns:

Type Description
str

A human-readable string, or empty string if no data is present.
Source code in SRToolkit/evaluation/result_augmentation.py 
544
545
546
547
548
549
550
551
552
553
554
555
556
@classmethod
def format_model_result(cls, data: Dict[str, Any]) -> str:
    """
    Format per-model BED data for display.

    Args:
        data: Augmentation dict containing ``"bed"``.

    Returns:
        A human-readable string, or empty string if no data is present.
    """
    val = data.get("bed", "")
    return f"BED={val}" if val != "" else ""

to_dict 

to_dict(base_path: str, name: str) -> dict

Creates a dictionary representation of the BED augmenter.

Parameters:

Name Type Description Default
base_path str

Used to save the data of the evaluator to disk.

 required
name str

Used to save the data of the evaluator to disk.

 required

Returns:

Type Description
dict

A dictionary containing the necessary information to recreate the augmenter.
Source code in SRToolkit/evaluation/result_augmentation.py 
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
def to_dict(self, base_path: str, name: str) -> dict:
    """
    Creates a dictionary representation of the BED augmenter.

    Args:
        base_path: Used to save the data of the evaluator to disk.
        name: Used to save the data of the evaluator to disk.

    Returns:
        A dictionary containing the necessary information to recreate the augmenter.
    """
    return {
        "format_version": 1,
        "name": self.name,
        "type": "BED",
        "scope": self.scope,
        "evaluator": self.evaluator.to_dict(base_path, name + "_BED_augmenter"),
    }

from_dict staticmethod 

from_dict(data: dict) -> BED

Creates an instance of the BED augmenter from a dictionary.

Parameters:

Name Type Description Default
data dict

A dictionary containing the necessary information to recreate the augmenter.

 required

Returns:

Type Description
BED

An instance of the BED augmenter.
Source code in SRToolkit/evaluation/result_augmentation.py 
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
@staticmethod
def from_dict(data: dict) -> "BED":
    """
    Creates an instance of the BED augmenter from a dictionary.

    Args:
        data: A dictionary containing the necessary information to recreate the augmenter.

    Returns:
        An instance of the BED augmenter.
    """
    if data.get("format_version", 1) != 1:
        raise ValueError(f"[BED.from_dict] Unsupported format_version: {data.get('format_version')!r}. Expected 1.")
    evaluator = SR_evaluator.from_dict(data["evaluator"])
    return BED(evaluator, scope=data["scope"], name=data["name"])

R2 

R2(evaluator: SR_evaluator, scope: str = 'top', name: str = 'R2')

Bases: ResultAugmenter

Computes R² for the top models using a separate evaluator (e.g. a held-out test set).

The same evaluator instance can be shared with RMSE to avoid loading test data twice.

Parameters:

Name Type Description Default
evaluator SR_evaluator

SR_evaluator used to score the models. Must be initialized with ranking_function="rmse" and a non-None y.

 required
scope str

Which expressions to score.

  • "best": only the best expression.
  • "top": the best expression and all top-k models.
  • "all": everything in "top" plus all evaluated expressions.
 'top'
name str

Key used in augmentations dict of EvalResult and ModelResult. Default "R2".

 'R2'

Raises:

Type Description
Exception

If evaluator.ranking_function != "rmse" or evaluator.y is None.
Source code in SRToolkit/evaluation/result_augmentation.py 
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
def __init__(self, evaluator: SR_evaluator, scope: str = "top", name: str = "R2") -> None:  # noqa: F821
    """
    Computes R² for the top models using a separate evaluator (e.g. a held-out test set).

    The same evaluator instance can be shared with
    [RMSE][SRToolkit.evaluation.result_augmentation.RMSE] to avoid loading test data twice.

    Args:
        evaluator: [SR_evaluator][SRToolkit.evaluation.sr_evaluator.SR_evaluator] used to
            score the models. Must be initialized with ``ranking_function="rmse"`` and a
            non-``None`` ``y``.
        scope: Which expressions to score.

            - ``"best"``: only the best expression.
            - ``"top"``: the best expression and all top-k models.
            - ``"all"``: everything in ``"top"`` plus all evaluated expressions.
        name: Key used in
            ``augmentations`` dict of [EvalResult][SRToolkit.utils.types.EvalResult] and
            [ModelResult][SRToolkit.utils.types.ModelResult].
            Default ``"R2"``.

    Raises:
        Exception: If ``evaluator.ranking_function != "rmse"`` or ``evaluator.y is None``.
    """
    super().__init__(name)

    if scope not in ["best", "top", "all"]:
        raise Exception(f"[R2 augmenter] Invalid scope: {scope}. Must be one of 'best', 'top', 'all'.")
    self.scope = scope

    self.evaluator = evaluator
    if self.evaluator.ranking_function != "rmse":
        raise Exception("[R2 augmenter] Ranking function of the evaluator must be set to 'rmse' to compute R^2.")
    if self.evaluator.y is None:
        raise Exception("[R2 augmenter] y in the evaluator must not be None to compute R^2.")
    self.ss_tot = np.sum((self.evaluator.y - np.mean(self.evaluator.y)) ** 2)

write_results 

write_results(results: EvalResult) -> None

Write R² scores into results and its models.

Stores {"best_expr_r^2": ...} in EvalResult augmentations and {"r^2": ..., "parameters_r^2": ...} in each model's augmentations when scope is "top" or "all".

Parameters:

Name Type Description Default
results EvalResult

The EvalResult to augment.

 required
Source code in SRToolkit/evaluation/result_augmentation.py 
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
def write_results(self, results: EvalResult) -> None:
    """
    Write R² scores into *results* and its models.

    Stores ``{"best_expr_r^2": ...}`` in
    [EvalResult][SRToolkit.utils.types.EvalResult] ``augmentations`` and
    ``{"r^2": ..., "parameters_r^2": ...}`` in each model's augmentations when ``scope``
    is ``"top"`` or ``"all"``.

    Args:
        results: The [EvalResult][SRToolkit.utils.types.EvalResult] to augment.
    """
    eval_data: Dict[str, Any] = {"best_expr_r^2": self._compute_r2(results.top_models[0])}
    results.add_augmentation(self.name, eval_data, self._type)

    if self.scope == "top" or self.scope == "all":
        for model in results.top_models:
            key = "".join(model.expr)
            top_model_data: Dict[str, Any] = {
                "r^2": self._compute_r2(model),
                "parameters_r^2": self.evaluator.models[key].parameters,
            }
            model.add_augmentation(self.name, top_model_data, self._type)

    if self.scope == "all":
        for model in results.all_models:
            key = "".join(model.expr)
            all_model_data: Dict[str, Any] = {
                "r^2": self._compute_r2(model),
                "parameters_r^2": self.evaluator.models[key].parameters,
            }
            model.add_augmentation(self.name, all_model_data, self._type)

format_eval_result classmethod 

format_eval_result(data: Dict[str, Any]) -> str

Format experiment-level R² data for display.

Parameters:

Name Type Description Default
data Dict[str, Any]

Augmentation dict containing "best_expr_r^2".

 required

Returns:

Type Description
str

A human-readable string, or empty string if no data is present.
Source code in SRToolkit/evaluation/result_augmentation.py 
670
671
672
673
674
675
676
677
678
679
680
681
682
@classmethod
def format_eval_result(cls, data: Dict[str, Any]) -> str:
    """
    Format experiment-level R² data for display.

    Args:
        data: Augmentation dict containing ``"best_expr_r^2"``.

    Returns:
        A human-readable string, or empty string if no data is present.
    """
    val = data.get("best_expr_r^2", "")
    return f"Test R²: {val}" if val != "" else ""

format_model_result classmethod 

format_model_result(data: Dict[str, Any]) -> str

Format per-model R² data for display.

Parameters:

Name Type Description Default
data Dict[str, Any]

Augmentation dict containing "r^2" and optionally "parameters_r^2".

 required

Returns:

Type Description
str

A human-readable string with R² and fitted parameters.
Source code in SRToolkit/evaluation/result_augmentation.py 
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
@classmethod
def format_model_result(cls, data: Dict[str, Any]) -> str:
    """
    Format per-model R² data for display.

    Args:
        data: Augmentation dict containing ``"r^2"`` and optionally ``"parameters_r^2"``.

    Returns:
        A human-readable string with R² and fitted parameters.
    """
    parts = [f"R²={data['r^2']:.4g}"]
    if "parameters_r^2" in data and data["parameters_r^2"] is not None:
        parts.append(f"params={np.round(data['parameters_r^2'], 4).tolist()}")
    return ", ".join(parts)

to_dict 

to_dict(base_path: str, name: str) -> dict

Creates a dictionary representation of the R2 augmenter.

Parameters:

Name Type Description Default
base_path str

Used to save the data of the evaluator to disk.

 required
name str

Used to save the data of the evaluator to disk.

 required

Returns:

Type Description
dict

A dictionary containing the necessary information to recreate the augmenter.
Source code in SRToolkit/evaluation/result_augmentation.py 
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
def to_dict(self, base_path: str, name: str) -> dict:
    """
    Creates a dictionary representation of the R2 augmenter.

    Args:
        base_path: Used to save the data of the evaluator to disk.
        name: Used to save the data of the evaluator to disk.

    Returns:
        A dictionary containing the necessary information to recreate the augmenter.
    """
    return {
        "format_version": 1,
        "name": self.name,
        "type": "R2",
        "scope": self.scope,
        "evaluator": self.evaluator.to_dict(base_path, name + "_R2_augmenter"),
    }

from_dict staticmethod 

from_dict(data: dict) -> R2

Creates an instance of the R2 augmenter from a dictionary.

Parameters:

Name Type Description Default
data dict

A dictionary containing the necessary information to recreate the augmenter.

 required

Returns:

Type Description
R2

An instance of the R2 augmenter.
Source code in SRToolkit/evaluation/result_augmentation.py 
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
@staticmethod
def from_dict(data: dict) -> "R2":
    """
    Creates an instance of the R2 augmenter from a dictionary.

    Args:
        data: A dictionary containing the necessary information to recreate the augmenter.

    Returns:
        An instance of the R2 augmenter.
    """
    if data.get("format_version", 1) != 1:
        raise ValueError(f"[R2.from_dict] Unsupported format_version: {data.get('format_version')!r}. Expected 1.")
    evaluator = SR_evaluator.from_dict(data["evaluator"])
    return R2(evaluator, scope=data["scope"], name=data["name"])