π 2023ΒΆ
Note
ID: cards.CFPB.product.2023
| Type: TaskCard
{
"loader": {
"files": {
"train": "https://www.consumerfinance.gov/data-research/consumer-complaints/search/api/v1/?date_received_max=2023-01-04&date_received_min=2022-01-04&field=all&format=csv&has_narrative=true&lens=product&no_aggs=true&size=340390&sub_lens=sub_product&trend_depth=5&trend_interval=month"
},
"type": "load_csv"
},
"preprocess_steps": [
{
"mix": {
"test": "train[20%]",
"train": "train[70%]",
"validation": "train[10%]"
},
"type": "split_random_mix"
},
{
"field_to_field": {
"Consumer complaint narrative": "text",
"Product": "label"
},
"type": "rename_fields"
},
{
"mappers": {
"label": {
"Checking or savings account": "checking or savings account",
"Credit card or prepaid card": "credit card or prepaid card",
"Credit reporting, credit repair services, or other personal consumer reports": "credit reporting or credit repair services or other personal consumer reports",
"Debt collection": "debt collection",
"Money transfer, virtual currency, or money service": "money transfer or virtual currency or money service",
"Mortgage": "mortgage",
"Payday loan, title loan, or personal loan": "payday loan or title loan or personal loan",
"Student loan": "student loan",
"Vehicle loan or lease": "vehicle loan or lease"
}
},
"type": "map_instance_values"
},
{
"fields": {
"classes": [
"credit reporting or credit repair services or other personal consumer reports",
"credit card or prepaid card",
"payday loan or title loan or personal loan",
"debt collection",
"mortgage",
"checking or savings account",
"money transfer or virtual currency or money service",
"vehicle loan or lease",
"student loan"
],
"text_type": "text",
"type_of_class": "topic"
},
"type": "add_fields"
}
],
"task": "tasks.classification.multi_class",
"templates": "templates.classification.multi_class.all",
"type": "task_card"
}
Explanation about TaskCardΒΆ
TaskCard delineates the phases in transforming the source dataset into a model-input, and specifies the metrics for evaluation of model-output.
- Attributes:
loader: specifies the source address and the loading operator that can access that source and transform it into a unitxt multistream.
preprocess_steps: list of unitxt operators to process the data source into a model-input.
task: specifies the fields (of the already (pre)processed instance) making the inputs, the fields making the outputs, and the metrics to be used for evaluating the model output.
templates: format strings to be applied on the input fields (specified by the task) and the output fields. The template also carries the instructions and the list of postprocessing steps, to be applied to the model output.
Explanation about RenameFieldsΒΆ
Renames fields.
Move value from one field to another, potentially, if βuse_queryβ=True, from one branch into another. Remove the from field, potentially part of it in case of use_query.
- Examples:
RenameFields(field_to_field={βbβ: βcβ}) will change inputs [{βaβ: 1, βbβ: 2}, {βaβ: 2, βbβ: 3}] to [{βaβ: 1, βcβ: 2}, {βaβ: 2, βcβ: 3}]
RenameFields(field_to_field={βbβ: βc/dβ}, use_query=True) will change inputs [{βaβ: 1, βbβ: 2}, {βaβ: 2, βbβ: 3}] to [{βaβ: 1, βcβ: {βdβ: 2}}, {βaβ: 2, βcβ: {βdβ: 3}}]
RenameFields(field_to_field={βbβ: βb/dβ}, use_query=True) will change inputs [{βaβ: 1, βbβ: 2}, {βaβ: 2, βbβ: 3}] to [{βaβ: 1, βbβ: {βdβ: 2}}, {βaβ: 2, βbβ: {βdβ: 3}}]
RenameFields(field_to_field={βb/c/eβ: βb/dβ}, use_query=True) will change inputs [{βaβ: 1, βbβ: {βcβ: {βeβ: 2, βfβ: 20}}}] to [{βaβ: 1, βbβ: {βcβ: {βfβ: 20}, βdβ: 2}}]
Explanation about SplitRandomMixΒΆ
Splits a multistream into new streams (splits), whose names, source input stream, and amount of instances, are specified by arg βmixβ.
The keys of arg βmixβ, are the names of the new streams, the values are of the form: βname-of-source-stream[percentage-of-source-stream]β Each input instance, of any input stream, is selected exactly once for inclusion in any of the output streams.
Examples: When processing a multistream made of two streams whose names are βtrainβ and βtestβ, by SplitRandomMix(mix = { βtrainβ: βtrain[99%]β, βvalidationβ: βtrain[1%]β, βtestβ: βtestβ }) the output is a multistream, whose three streams are named βtrainβ, βvalidationβ, and βtestβ. Output stream βtrainβ is made of randomly selected 99% of the instances of input stream βtrainβ, output stream βvalidationβ is made of the remaining 1% instances of input βtrainβ, and output stream βtestβ is made of the whole of input stream βtestβ.
When processing the above input multistream by SplitRandomMix(mix = { βtrainβ: βtrain[50%]+test[0.1]β, βvalidationβ: βtrain[50%]+test[0.2]β, βtestβ: βtest[0.7]β }) the output is a multistream, whose three streams are named βtrainβ, βvalidationβ, and βtestβ. Output stream βtrainβ is made of randomly selected 50% of the instances of input stream βtrainβ + randomly selected 0.1 (i.e., 10%) of the instances of input stream βtestβ. Output stream βvalidationβ is made of the remaining 50% instances of input βtrainβ+ randomly selected 0.2 (i.e., 20%) of the original instances of input βtestβ, that were not selected for output βtrainβ, and output stream βtestβ is made of the remaining instances of input βtestβ.
Explanation about MapInstanceValuesΒΆ
A class used to map instance values into other values.
This class is a type of StreamInstanceOperator, it maps values of instances in a stream using predefined mappers.
- Attributes:
- mappers (Dict[str, Dict[str, str]]): The mappers to use for mapping instance values.
Keys are the names of the fields to be mapped, and values are dictionaries that define the mapping from old values to new values.
- strict (bool): If True, the mapping is applied strictly. That means if a value
does not exist in the mapper, it will raise a KeyError. If False, values that are not present in the mapper are kept as they are.
- process_every_value (bool): If True, all fields to be mapped should be lists, and the mapping
is to be applied to their individual elements. If False, mapping is only applied to a field containing a single value.
- Examples:
MapInstanceValues(mappers={βaβ: {β1β: βhiβ, β2β: βbyeβ}}) replaces β1β with βhiβ and β2β with βbyeβ in field βaβ in all instances of all streams: instance {βaβ:β1β, βbβ: 2} becomes {βaβ:βhiβ, βbβ: 2}.
MapInstanceValues(mappers={βaβ: {β1β: βhiβ, β2β: βbyeβ}}, process_every_element=True) Assuming field βaβ is a list of values, potentially including β1β-s and β2β-s, this replaces each such β1β with βhiβ and β2β β with βbyeβ in all instances of all streams: instance {βaβ: [β1β, β2β], βbβ: 2} becomes {βaβ: [βhiβ, βbyeβ], βbβ: 2}.
MapInstanceValues(mappers={βaβ: {β1β: βhiβ, β2β: βbyeβ}}, strict=True) To ensure that all values of field βaβ are mapped in every instance, use strict=True. Input instance {βaβ:β3β, βbβ: 2} will raise an exception per the above call, because β3β is not a key in the mapper of βaβ.
MapInstanceValues(mappers={βaβ: {str([1,2,3,4]): βAllβ, str([]): βNoneβ}}, strict=True) replaces a list [1,2,3,4] with the string βAllβ and an empty list by string βNoneβ. Note that mapped values are defined by their string representation, so mapped values must be converted to strings.
Explanation about AddFieldsΒΆ
Adds specified fields to each instance in a given stream or all streams (default) If fields exist, updates them.
- Args:
fields (Dict[str, object]): The fields to add to each instance. use_query (bool) : Use β/β to access inner fields use_deepcopy (bool) : Deep copy the input value to avoid later modifications
- Examples:
# Add a βclassesβ field with a value of a list βpositiveβ and βnegativeβ to all streams AddFields(fields={βclassesβ: [βpositiveβ,βnegativesβ]})
# Add a βstartβ field under the βspanβ field with a value of 0 to all streams AddFields(fields={βspan/startβ: 0}
# Add a βclassesβ field with a value of a list βpositiveβ and βnegativeβ to βtrainβ stream AddFields(fields={βclassesβ: [βpositiveβ,βnegativesβ], apply_to_stream=[βtrainβ]})
# Add a βclassesβ field on a given list, prevent modification of original list # from changing the instance. AddFields(fields={βclassesβ: alist}), use_deepcopy=True) # if now alist is modified, still the instances remain intact.
References: templates.classification.multi_class.all, tasks.classification.multi_class
Read more about catalog usage here.