model is not fitted yet!

[LiuHao-THU]

🐛 Bug Description

When I Run Prediction, I met This problem【model is not fitted yet! 】

I run this code in jupyter lab, When I run this code for the second time, The code runs correctly.

To Reproduce

Steps to reproduce the behavior:

Here is the code.

from qlib.workflow.record_temp import SignalRecord, PortAnaRecord, SigAnaRecord

rid = '61cc4cd94ae24137ac196d7e82c8e666'

# backtest and analysis
with R.start(experiment_name='PatchTsT', recorder_id=rid, resume=True): #
    # get model
    model = R.load_object("trained_model")
    # signal-based analysis
    rec = R.get_recorder()
    sar = SigAnaRecord(rec)
    sar.generate()

    # prediction
    sr = SignalRecord(model, dataset, rec)
    sr.generate()
    ba_rid = rec.id 
    
    #  portfolio-based analysis: backtest
    par = PortAnaRecord(rec, port_analysis_config, "day")
    par.generate()

[LiuHao-THU]

The previous link That Mentions This Problem:

#1340

[LiuHao-THU]

solution to this problem is than the config should in recorder

rid = '5fb97c7230cd414abcfaa382585a2474'

with R.start(experiment_name="ALSTM_Final", recorder_id=rid, resume=True):
#     R.log_params("depth", 3)

    recorder = R.get_recorder(recorder_id=rid, experiment_name="ALSTM_Final")
    
    model = recorder.load_object("trained_model")
    
    port_analysis_config = {
        "executor": {
            "class": "SimulatorExecutor",
            "module_path": "qlib.backtest.executor",
            "kwargs": {
                "time_per_step": "day",
                "generate_portfolio_metrics": True,
            },
        },
        "strategy": {
            "class": "TopkDropoutStrategy",
            "module_path": "qlib.contrib.strategy.signal_strategy",
            "kwargs": {
                "model": model,
                "dataset": dataset,
                "topk":50,
                "n_drop": 5,
            },
        },
        "backtest": {
            "start_time": "2019-01-01",
            "end_time": "2023-04-01",
            "account": 100000000,
            "benchmark": benchmark,
            "exchange_kwargs": {
                "freq": "day",
                "limit_threshold": 0.095,
                "deal_price": "close",
                "open_cost": 0.0005,
                "close_cost": 0.0015,
                "min_cost": 5,
            },
        },
    }    
    
    # prediction
    sr = SignalRecord(model, dataset, recorder)
    sr.generate()    
    ba_rid = recorder.id    

    # signal-based analysis
    rec = R.get_recorder()
    sar = SigAnaRecord(recorder)
    sar.generate()
    
#      portfolio-based analysis: backtest
    par = PortAnaRecord(recorder, port_analysis_config, "day")
    par.generate()