What causes "astype not supported: {astype}" in microsoft/qlib?

Use astype='gbdt' or astype='linear' exactly (lowercase)

microsoft/qlib · error · ValueError

astype not supported: {astype}

Error message

astype not supported: {astype}

What it means

DDGDA._adjust_task (qlib/contrib/rolling/ddgda.py) reconfigures the base task for either a GBDT model (dropping processors) or a linear model (adding PROC_ARGS). The astype argument must be exactly 'gbdt' or 'linear'; anything else raises ValueError('astype not supported: ...').

Source

Thrown at qlib/contrib/rolling/ddgda.py:156

        """
        # NOTE: here is just for aligning with previous implementation
        # It is not necessary for the current implementation
        handler = task["dataset"].setdefault("kwargs", {}).setdefault("handler", {})
        if astype == "gbdt":
            task["model"] = LGBM_MODEL
            if isinstance(handler, dict):
                # We don't need preprocessing when using GBDT model
                for k in ["infer_processors", "learn_processors"]:
                    if k in handler.setdefault("kwargs", {}):
                        handler["kwargs"].pop(k)
        elif astype == "linear":
            task["model"] = LINEAR_MODEL
            if isinstance(handler, dict):
                handler["kwargs"].update(PROC_ARGS)
            else:
                self.logger.warning("The handler can't be adjusted.")
        else:
            raise ValueError(f"astype not supported: {astype}")
        return task

    def _get_feature_importance(self):
        # this must be lightGBM, because it needs to get the feature importance
        task = self.basic_task(enable_handler_cache=False)
        task = self._adjust_task(task, astype="gbdt")
        task = replace_task_handler_with_cache(task, self.working_dir)

        with R.start(experiment_name="feature_importance"):
            model = init_instance_by_config(task["model"])
            dataset = init_instance_by_config(task["dataset"])
            model.fit(dataset)

        fi = model.get_feature_importance()
        # Because the model use numpy instead of dataframe for training lightgbm
        # So the we must use following extra steps to get the right feature importance
        df = dataset.prepare(segments=slice(None), col_set="feature", data_key=DataHandlerLP.DK_R)
        cols = df.columns

View on GitHub (pinned to 79633dd950)

Solutions

Use astype='gbdt' or astype='linear' exactly (lowercase)
If you subclass DDGDA, override _adjust_task to handle your custom astype values before delegating to super()
Check the value coming from your workflow config for typos or case differences

Example fix

# before
task = self._adjust_task(task, astype='GBDT')

# after
task = self._adjust_task(task, astype='gbdt')

Defensive patterns

Strategy: validation

Validate before calling

assert astype in ('gbdt', 'linear'), f'astype must be gbdt or linear, got {astype!r}'
task = self._adjust_task(task, astype=astype)

Type guard

def is_supported_astype(astype: str) -> bool:
    return astype in {'gbdt', 'linear'}

Prevention

Lowercase and whitelist astype before calling _adjust_task
Subclasses adding model families should extend the accepted set inside their own override

When it happens

Trigger: Calling _adjust_task(task, astype=...) with a value other than 'gbdt' or 'linear', e.g. 'lstm', 'GBDT' (case-sensitive), or None. Internal call sites use astype='gbdt' for feature importance and the configured type elsewhere.

Common situations: Subclassing DDGDA and extending _adjust_task with a new model family without handling the new astype string; passing an uppercase or misspelled model type from a custom config.

Related errors

AI-assisted analysis of microsoft/qlib@79633dd950 (2026-08-15). Data as JSON: /api/errors/7e58c56d714fbf52. Report an issue: GitHub.