from onyxengine.data import OnyxDataset
dataset = OnyxDataset(
dataframe: pd.DataFrame = pd.DataFrame(),
features: List[str] = [],
dt: float = 0,
config: OnyxDatasetConfig = None
)
Parameters
dataframe
pd.DataFrame
default:"pd.DataFrame()"
The pandas DataFrame containing the time series data.
List of column names to include as features.
Time step between samples in seconds.
config
OnyxDatasetConfig
default:"None"
Optional configuration object. If provided, features and dt are ignored.
Attributes
The underlying pandas DataFrame.
Configuration object containing:
features: List of feature namesdt: Time step in secondstype: Always “dataset”
Example
Create from DataFrame
import pandas as pd
from onyxengine.data import OnyxDataset
# Load your data
df = pd.read_csv('sensor_data.csv')
# Create dataset
dataset = OnyxDataset(
dataframe=df,
features=['acceleration', 'velocity', 'position', 'control'],
dt=0.01 # 100 Hz sampling
)
# Access properties
print(dataset.dataframe.shape)
print(dataset.config.features)
print(dataset.config.dt)
Load from Engine
from onyxengine import Onyx
# Initialize the client
onyx = Onyx()
# Datasets returned by load_dataset are OnyxDataset objects
dataset = onyx.load_dataset('example_train_data')
print(type(dataset)) # <class 'onyxengine.data.dataset.OnyxDataset'>
print(dataset.config.features)
print(dataset.config.dt)
OnyxDatasetConfig
The config object contains metadata about the dataset:
from onyxengine.data import OnyxDatasetConfig
config = OnyxDatasetConfig(
features=['accel', 'vel', 'pos'],
dt=0.01
)
| Attribute | Type | Description |
|---|
type | str | Always “dataset” (read-only) |
features | List[str] | Feature column names |
dt | float | Time step in seconds |
- Contain numeric columns for all features
- Have no missing values (NaN)
- Be sampled at regular time intervals
- Have columns matching the feature names
# Prepare data
df = df[features].dropna()
for col in df.columns:
df[col] = df[col].astype('float32')
dataset = OnyxDataset(dataframe=df, features=features, dt=0.01)
