Setting up an MLflow server locally is simple. Use the following command:
mlflow server --host 127.0.0.1 --port 8080
Then set the tracking URI.
mlflow.set_tracking_uri("http://127.0.0.1:8080")
For more advanced settings, see MLflow documentation.
For this article, we use the California Housing Dataset (CC BY license). However, you can apply the same principles to record and track any data set of your choice.
For more information on the California housing data set, see this doctor.
mlflow.data.dataset.Dataset
Before delving into recording, evaluating, and retrieving data sets, it is important to understand the concept of data sets in MLflow. MLflow provides the mlflow.data.dataset.Dataset object, which represents data sets used with MLflow Tracking.
class mlflow.data.dataset.Dataset(source: mlflow.data.dataset_source.DatasetSource, name: Optional(str) = None, digest: Optional(str) = None)
This object comes with key properties:
- A required parameter, fountain (the data source of your dataset like
mlflow.data.dataset_source.DatasetSourceobject) - digest (fingerprint for your data set) and name (name of your dataset), which can be set via parameters.
- scheme and profile to describe the structure of the data set and the statistical properties.
- Information about the data set. fountainas its storage location.
You can easily convert the data set into a dictionary using to_dict() or a JSON string using to_json().
Support for popular data set formats
MLflow makes it easy to work with various types of data sets through specialized classes that extend the core. mlflow.data.dataset.Dataset. As of this writing, these are some of the notable dataset classes supported by MLflow:
- pandas:
mlflow.data.pandas_dataset.PandasDataset - NumPy:
mlflow.data.numpy_dataset.NumpyDataset - Spark:
mlflow.data.spark_dataset.SparkDataset - hugging face:
mlflow.data.huggingface_dataset.HuggingFaceDataset - TensorFlow:
mlflow.data.tensorflow_dataset.TensorFlowDataset - Evaluation Data Sets:
mlflow.data.evaluation_dataset.EvaluationDataset
All of these classes come with a convenient mlflow.data.from_* API to load datasets directly into MLflow. This makes it easy to build and manage data sets, regardless of their underlying format.
mlflow.data.dataset_source.DatasetSource
He mlflow.data.dataset.DatasetSource The class is used to represent the source of the dataset in MLflow. When creating a mlflow.data.dataset.Dataset object, the source The parameter can be specified as a string (for example, a file path or URL) or as an instance of the mlflow.data.dataset.DatasetSource class.
class mlflow.data.dataset_source.DatasetSource
If a string is provided like sourceMLflow internally calls resolve_dataset_source function. This function iterates through a predefined list of data sources and DatasetSource classes to determine the most appropriate type of font. However, MLflow's ability to accurately resolve the source of the data set is limited, especially when the candidate_sources The argument (a list of potential sources) is set to Nonewhich is the default value.
In cases where the DatasetSource The class cannot resolve the raw source, an MLflow exception is raised. For best practices, I recommend explicitly creating and using an instance of the mlflow.data.dataset.DatasetSource class when defining the origin of the data set.
class HTTPDatasetSource(DatasetSource)class DeltaDatasetSource(DatasetSource)class FileSystemDatasetSource(DatasetSource)class HuggingFaceDatasetSource(DatasetSource)class SparkDatasetSource(DatasetSource)
One of the easiest ways to register data sets in MLflow is through the mlflow.log_input() API. This allows you to record data sets in any format that is supported by mlflow.data.dataset.Datasetwhich can be extremely useful when managing large-scale experiments.
Step by step guide
First, let's find the California Housing data set and convert it to a pandas.DataFrame for easier handling. Here, we create a data frame that combines the feature data (california_data) and the destination data (california_target).
california_housing = fetch_california_housing()
california_data: pd.DataFrame = pd.DataFrame(california_housing.data, columns=california_housing.feature_names)
california_target: pd.DataFrame = pd.DataFrame(california_housing.target, columns=('Target'))california_housing_df: pd.DataFrame = pd.concat((california_data, california_target), axis=1)
To register the dataset with meaningful metadata, we define some parameters such as data source URL, dataset name, and target column. These will provide useful context when we retrieve the data set later.
If we look deeper into the fetch_california_housing source codewe can see that the data originated from https://www.dcc.fc.up.pt/~ltorgo/Regression/cal_housing.tgz.
dataset_source_url: str = 'https://www.dcc.fc.up.pt/~ltorgo/Regression/cal_housing.tgz'
dataset_source: DatasetSource = HTTPDatasetSource(url=dataset_source_url)
dataset_name: str = 'California Housing Dataset'
dataset_target: str = 'Target'
dataset_tags = {
'description': california_housing.DESCR,
}
Once the data and metadata are defined, we can convert the pandas.DataFrame in a mlflow.data.Dataset object.
dataset: PandasDataset = mlflow.data.from_pandas(
df=california_housing_df, source=dataset_source, targets=dataset_target, name=dataset_name
)print(f'Dataset name: {dataset.name}')
print(f'Dataset digest: {dataset.digest}')
print(f'Dataset source: {dataset.source}')
print(f'Dataset schema: {dataset.schema}')
print(f'Dataset profile: {dataset.profile}')
print(f'Dataset targets: {dataset.targets}')
print(f'Dataset predictions: {dataset.predictions}')
print(dataset.df.head())
Example output:
Dataset name: California Housing Dataset
Dataset digest: 55270605
Dataset source:
Dataset schema: ('MedInc': double (required), 'HouseAge': double (required), 'AveRooms': double (required), 'AveBedrms': double (required), 'Population': double (required), 'AveOccup': double (required), 'Latitude': double (required), 'Longitude': double (required), 'Target': double (required))
Dataset profile: {'num_rows': 20640, 'num_elements': 185760}
Dataset targets: Target
Dataset predictions: None
MedInc HouseAge AveRooms AveBedrms Population AveOccup Latitude Longitude Target
0 8.3252 41.0 6.984127 1.023810 322.0 2.555556 37.88 -122.23 4.526
1 8.3014 21.0 6.238137 0.971880 2401.0 2.109842 37.86 -122.22 3.585
2 7.2574 52.0 8.288136 1.073446 496.0 2.802260 37.85 -122.24 3.521
3 5.6431 52.0 5.817352 1.073059 558.0 2.547945 37.85 -122.25 3.413
4 3.8462 52.0 6.281853 1.081081 565.0 2.181467 37.85 -122.25 3.422
Note that you can even convert the data set to a dictionary to access additional properties like source_type:
for k,v in dataset.to_dict().items():
print(f"{k}: {v}")
name: California Housing Dataset
digest: 55270605
source: {"url": "https://www.dcc.fc.up.pt/~ltorgo/Regression/cal_housing.tgz"}
source_type: http
schema: {"mlflow_colspec": ({"type": "double", "name": "MedInc", "required": true}, {"type": "double", "name": "HouseAge", "required": true}, {"type": "double", "name": "AveRooms", "required": true}, {"type": "double", "name": "AveBedrms", "required": true}, {"type": "double", "name": "Population", "required": true}, {"type": "double", "name": "AveOccup", "required": true}, {"type": "double", "name": "Latitude", "required": true}, {"type": "double", "name": "Longitude", "required": true}, {"type": "double", "name": "Target", "required": true})}
profile: {"num_rows": 20640, "num_elements": 185760}
Now that we have our dataset ready, it's time to log it to an MLflow run. This allows us to capture the metadata of the dataset, making it part of the experiment for future reference.
with mlflow.start_run():
mlflow.log_input(dataset=dataset, context='training', tags=dataset_tags)
View run sassy-jay-279 at: http://127.0.0.1:8080/#/experiments/0/runs/5ef16e2e81bf40068c68ce536121538c
View experiment at: http://127.0.0.1:8080/#/experiments/0
Let's explore the dataset in the MLflow() UI. You will find your data set in the default experiment list. In it Data sets used In the section, you can see the context of the dataset, which in this case is marked as used for training. Additionally, all relevant fields and properties of the data set will be displayed.
Congratulations! You have recorded your first set of data!
