Quick Start Guide

This guide walks you through a simple usage example of the ts-shape library to load, transform, and analyze time series data.

Installation

You can install the library via pip:

pip install ts-shape

Example Workflow

1. Import Modules

from ts_shape.loader.timeseries import parquet_loader
from ts_shape.transform.filter import boolean_filter
from ts_shape.transform.time_functions import timestamp_converter, timezone_shift
from ts_shape.features.stats import string_stats

2. Load Time Series Data

Load all parquet files from a given directory:

base_path = 'path/to/your/parquet/files'
df_all = parquet_loader.ParquetLoader.load_all_files(base_path)

3. Filter Data

Filter rows where the column is_delta is True:

df_is_delta = boolean_filter.IsDeltaFilter.filter_is_delta_true(df_all)

4. Convert Timestamps

Convert Unix nanosecond timestamps to timezone-aware datetime objects:

df_timestamp = timestamp_converter.TimestampConverter.convert_to_datetime(
    dataframe=df_is_delta,
    columns=['systime', 'plctime'],
    unit='ns',
    timezone='UTC'
)

5. Shift Timezone

Convert timestamps from UTC to local time (e.g., Europe/Berlin):

df_timestamp_shift = timezone_shift.TimezoneShift.shift_timezone(
    dataframe=df_timestamp,
    time_column='systime',
    input_timezone='UTC',
    target_timezone='Europe/Berlin'
)

---

Next Steps

• Explore feature extraction with ts_shape.features.
• Chain multiple transformations into a pipeline.