python-tide 0.2.0

Pipeline Development and Data Visualization for Time Series in Physical Measurements

Tide is a tool for developing data processing pipelines and visualizing time series data, particularly suited for physical measurements. Key features include:

• Efficient Data Management

  • Organize and select data using a tagging system

• Pipeline Construction

  • Store and retrieve pipelines easily with JSON-based dictionary structures
  • Build dynamic pipelines that adjust based on the selected data

• Interactive Visualization

  • Create interactive plots to explore data (plotly)
  • Visualize pipeline or slices of pipelines effects on data

• Custom Data Enrichment

  • Integrate external weather data sources
  • Implement autoregressive models for gaps filling
  • Develop and incorporate custom data processors

Uses pandas DataFrames and Series for robust data handling. bigtree for tags and data selection. Scikit-learn's API for pipeline construction.

Getting started

1- Install ⬇️

pip install python-tide

2- Load and format data 🌲

To begin, load your time series data into a pandas DataFrame, ensuring the index is a DateTimeIndex:

df = pd.read_csv(
    "https://raw.githubusercontent.com/BuildingEnergySimulationTools/tide/main/tutorials/getting_started_ts.csv",
    parse_dates=True,
    index_col=0
)

Rename columns using Tide's tagging system. The format is: name__unit__bloc__sub_bloc with tags separated by double underscores. The order of the tags matters. The order of tags is important, and you can use "OTHER" as a placeholder You can use one or several tags.

df.columns = ["Tin__°C__Building", "Text__°C__Outdoor", "Heat__W__Building"]

Plumber objects are used to help us with pipelines building and data visualization

from tide.plumbing import Plumber

plumber = Plumber(df)

Display the data organization as a tree:

plumber.show()

Select data using tags:

plumber.get_corrected_data("°C")
plumber.get_corrected_data("Building")
plumber.get_corrected_data("Tin")

3- Visualizing data 📈

Show data availability:

plumber.plot_gaps_heatmap(time_step='d')

Plot time series with missing data highlighted:

fig = plumber.plot(plot_gaps=True)
fig.show(None)

4- Building and testing Pipelines 🛠️

Create a pipeline dictionary:

pipe_dict = {
    "step_1": [["Common_proc_1"], ["Common_proc_2", ["arg1", "arg2"]]],
    "step_2": {
        "selection_1": [["Proc_selection_1", {"arg": "arg_value"}]]
    }
}

Pipeline Rules:

• Use dictionaries for pipeline description
• Keys represent pipeline steps ex. "step_1"
• Step values can be lists (apply to all columns) or dicts (filter columns)
• Processing objects are listed as [class_name, arguments]

Example Pipeline:

• Resample data to 15-minute intervals
• Interpolate temperature gaps ≤ 3 hours
• Fill large Tin gaps using Autoregressive STLForecast

pipe_dict = {
    "resample_15min": [["Resample", ["15min"]]],
    "interpolate_temps": {
        "°C": [["Interpolate", {"gaps_lte": "3h"}]]
    },
    "ar_tin": {
        "Tin": [
            [
                "FillGapsAR",
                {
                    "model_name": "Prophet",
                    "resample_at_td": "1h",
                    "gaps_gte": "3h",
                    "gaps_lte": "3d"
                }
            ]
        ]
    }
}

plumber.pipe_dict = pipe_dict

Get pipeline using get_pipeline method.

plumber.get_pipeline(verbose=True)

Get pipelines for specific columns

plumber.get_pipeline(select="Building", verbose=True)

Visualize pipeline effects:

plumber.plot(
    steps=None,
    plot_gaps=True,
    steps_2=slice(None, "interpolate_temps"),
    plot_gaps_2=True,
    verbose=True
)

Step Arguments:

• None: No operation (Identity)
• str: Process until named step
• list[str]: Perform specified steps
• slice: Process a slice of the pipeline

Compare full pipeline to raw data:

plumber.plot(
    steps=None,
    plot_gaps=True,
    steps_2=slice(None),
    plot_gaps_2=True,
    verbose=True
)

Sponsors

The development of this library has been supported by ENSNARE Project, which has received funding from the European Union's Horizon 2020 Research and Innovation Programme under Grant Agreement No. 953193. The sole responsibility for the content of this library lies entirely with the author's view. The European Commission is not responsible for any use that may be made of the information it contains.