MATLAB SDK

Get data from Marple DB in MATLAB

MATLAB allows for complex custom analysis on data that is stored in Marple DB

Installation

1. Get the latest matlab/ folder from the Marple SDK repository (or pin a release tag matlab-v*).

2. Add that folder to your MATLAB path:

addpath(genpath('/path/to/marple-sdk/matlab'));

1. Create config.json in the same directory as DB.m:

{
    // might be a different URL if you have self-hosted or VPC Marple
    "api_url": "https://db.marpledata.com/api/v1",
    "api_key": "<your-api-token-here>",
    "workspace": "<name of your workspace>"
}

1. Create the client:

mdb = DB.from_config();
% or: mdb = DB(api_url, api_key);

You can use MATLAB Online for testing (free tier available).

Usage

mdb = DB.from_config();

% list streams (always a cell array)
mdb.get_streams()

% list datasets in a stream (includes metadata)
mdb.get_datasets('flight data')

% download a signal as a table (time in nanoseconds)
mdb.get_data('2025-11-04-cessna-looping-test.tdms', 'altitude')

get_data returns a MATLAB table with a time column and a column named after the signal. Time is nanoseconds (UTC when the source used datetimes).

Upload a file and wait for import

dataset = mdb.push_file('flight data', 'run.csv', Metadata=struct('pilot', 'Ada'));
mdb.wait_for_import('flight data', dataset.id, Timeout=60);
T = mdb.get_data(dataset.path, 'altitude');

Use Overwrite=true to replace an existing dataset with the same name. For MATLAB .mat files, create a files stream with Plugin="mat".

Example output from mdb.get_data()

Add signals (custom ingest)

ds = mdb.add_dataset('flight data', 'custom-run');
% table with int64 ns time + value and/or value_text, or a timetable
mdb.add_signal('flight data', ds.id, 'altitude', my_table, Overwrite=false);

Upload returns after the API accepts completion; the Iceberg commit may still be running, so the signal may not be readable immediately.

Other helpers

• create_stream(name, ...) — create a stream (Type, Plugin, PluginArgs, …)

• get_signals(stream_name, dataset_id) — list signals

• update_metadata(stream_name, dataset_id, metadata) — merge dataset metadata

• clear_cache() — delete the local parquet cache

• health() — API health check

See DB.m and matlab/example.m in the repository for the full surface. Version history: matlab/CHANGELOG.md.

Resampling / combining signals

Combine tables with built-in MATLAB functions, for example:

data = mdb.get_data('2025-11-04-cessna-looping-test.tdms', 'altitude');

t1 = datetime(data.time/1e9, 'ConvertFrom', 'posixtime', 'TimeZone', 'UTC');
TT = table2timetable(table(t1, data.altitude, 'VariableNames', {'time', 'altitude'}), 'RowTimes', 'time');
TT_altitude = retime(TT, 'regular', 'nearest', 'TimeStep', seconds(2));

TT_all_data = synchronize(TT_altitude, TT_speed, TT_battery, 'union', 'nearest');

Caching

Downloaded parquet files are cached under, call mdb.clear_cache() to remove the folder

_marplecache/<workspace>/<datapool>/dataset=<id>/signal=<id>/

Compatibility (older MATLAB)

MATLAB older than R2023a cannot read ZSTD-compressed Parquet. On first use the SDK downloads a small parquet-transcode helper into matlab/_marplecache/ (Windows x64, macOS ARM, Linux x64). The same helper is used when uploading via add_signal.

Discover the MATLAB SDK in 1 minute

Want to discover more Marple features in 1 minute? Check out other 1 Minute Marple videos