Metadata

The class TimeSeriesDataset handles the metadata and some metrics from the loaded data. Information regarding the size of each dataset, before any of the preprocessing, as well as resamples and joined length of the dataset is generated here. This information can be retrived with function get_metadata().

Path	Description/Unit
x_hist	Histogram information for each tag. String with serialized JSON dictionary where key contains gap (<from>, <to>], and value number of samples within this gap.
data_provider	Data provider specific metadata. Result of GordoBaseDataProvider.get_metadata() method.
row_filter_tags	List of row filter tags. Tags participating in Row filter.
filtered_periods	Periods dropped by applied algorithm in the data preprocessing. Key is a filter periods type, and value is a list with drop_start, drop_end items in ISO timestamps.
summary_statistics	Descriptive statistics (quartiles, max/min, median etc.) for each tag
tag_loading_metadata.tags	Serialized tags information.
tag_loading_metadata.aggregate_metadata	DataFrame sizes: joined_length - after joining all columns, dropped_na_length - after removing missing values.

tag_loading_metadata also contains per tag metadata:

Path	Description/Unit
gaps	Empty time-series gaps. start, end of gaps in Unix-timestamp.
last_timestamp	First value in Unix-timestamp.
first_timestamp	Last value in Unix-timestamp.
original_length	Number of values.
resampled_length	Number of values after resampling.