SaschaHeyer

{
 "predictions": [
   {
     "confidence": 0.9409326314926147,
     "sentiment": "POSITIVE"
   }
 ]
}

SaschaHeyer

training_data_artifact = aiplatform.Artifact.create(
        schema_title='system.Dataset',
        uri='gs://doit-vertex-demo/higgs/training.csv',
        display_name='data')

with aiplatform.start_execution(
       schema_title="system.ContainerExecution", 
       display_name='training'
    ) as execution:

SaschaHeyer

experiment_run = aiplatform.ExperimentRun(
        run_name='run-1668456760',
        experiment='experiment-demo',
)

print(experiment_run.get_artifacts())
print(experiment_run.get_metrics())
print(experiment_run.get_params())
print(experiment_run.get_time_series_data_frame())
print(experiment_run.get_classification_metrics())

SaschaHeyer

experiment_df = aiplatform.get_experiment_df()
experiment_df = experiment_df[experiment_df.experiment_name == 'experiment-demo']
experiment_df

SaschaHeyer

aiplatform.init(
       project='sascha-playground-doit',
       location='us-central1',
       experiment='experiment-sample',
       experiment_tensorboard='projects/sascha-playground-doit/locations/us-central1/tensorboards/6382621018774568960'
    )

SaschaHeyer

experiment_tensorboard = vertex_ai.Tensorboard.create()

SaschaHeyer

training_data_artifact = aiplatform.Artifact.create(
        schema_title='system.Dataset',
        uri='gs://doit-vertex-demo/higgs/training.csv',
        display_name='training data')

SaschaHeyer

aiplatform.log_time_series_metrics({"mse": 2500.00, "rmse": 50.00})
aiplatform.log_time_series_metrics({"mse": 2600.00, "rmse": 60.00})
aiplatform.log_time_series_metrics({"mse": 2700.00, "rmse": 70.00})

SaschaHeyer

metrics = {}
metrics["f1"] = 0.81
aiplatform.log_metrics(metrics)

SaschaHeyer

from sklearn.metrics import confusion_matrix

labels = ["ant", "bird", "cat"]
y_true = ["cat", "ant", "cat", "cat", "ant", "bird"]
y_pred = ["ant", "ant", "cat", "cat", "ant", "cat"]

classification_metrics = {
        "matrix": confusion_matrix(y_true, y_pred, labels=labels).tolist(),
        "labels": labels
        }