Tutorial: Running Analysis Workflows

Choose analysis modules for a dataset, set their parameters, and start a pipeline run. This example uses the NDVI Index module to calculate vegetation-index statistics from multispectral orthomosaics or plot images.

There are two ways to start a run:

• Home → Set up an analysis: a guided flow that starts from your sensors, uploads the data, and offers only the modules that can read it.
• Editor → Run Analysis: a three-step wizard for a dataset that already holds data, using the modules installed in your account.

Both create a pipeline run that you follow on the Jobs page.

Before you start

You need a project with a dataset you can access, or the files to create one, and a running worker. Modules come from the shared catalog; an administrator must publish a module before anyone can select it.

NDVI needs red and near-infrared bands. Check the one-based NIR and red band indices against your sensor's band order before running. RGB imagery alone does not supply a near-infrared band.

Which modules can read my data?

Every dataset file has a modality (the sensor, such as multispectral) and a data product (what the file is, such as orthomosaic). Every module declares the pairs it reads as supported inputs; NDVI Index declares multispectral:orthomosaic and multispectral:image. A module can process a dataset when at least one of its pairs matches the dataset's files.

The Home guide applies this check for you. In the Editor wizard, compare the module's Supported inputs on its Modules page with the dataset's files before you select it. During a run, item-level modules skip plots and files with no matching input.

Option A: the guided setup on Home

Open Home and follow Set up an analysis. The guide has seven steps:

1. Platform: how the data was collected: proximal (ground), aerial, or satellite.
2. Sensors: the sensors you used. For each sensor, select Choose data products and check the formats you have. The step reports how many analyses are available for your data.
3. Project: under New project, enter a project, study, and dataset name; or pick them under Existing project. Select Create and continue.
4. Upload: add the files as a single file, a ZIP organized by plot, or a field-wide map with plot boundaries. See Importing Data.
5. Analyses: the list is filtered to modules that support your sensors and data formats. The Input filters panel shows each pair, such as Multispectral · Orthomosaic; pairs with no uploaded files appear dashed. Search for a goal, such as "vegetation index", and select Select on NDVI Index. Set Nir Band and Red Band, then select Add analysis. Selected modules appear as numbered steps under Selected analyses; use Parameters to change them later.
6. Review: check the inputs, analyses, and parameters, then tick I confirm these inputs, analyses, and parameters.
7. Run: select Run analysis.

Modules marked Not installed are installed in your account when the run starts. The run sends only the files matching the selected input pairs to the pipeline.

When the run is submitted, the guide offers Follow progress in Jobs, Visualize your data in the Editor, and Ask the Agent about your data.

Option B: Run Analysis from the Editor

Use this route for a dataset that already contains files.

1. Install the module

Open Modules, find NDVI Index under Discover, and select Install. The wizard lists only modules installed in your account. See Manage LgoPy Modules for versions, citations, and supported inputs.

2. Choose the functions

Open Editor, right-click the dataset in the data tree, and select Run Analysis. The dialog opens on Functions with the dataset already selected. (When the dataset is not preselected, the step is called Dataset & Functions and asks for the Project, Study, and Select Dataset.)

Under Available Functions, search or pick a Category, then check NDVI Index. Select Next.

3. Order the steps

Order lists the selected modules. Use the arrows to change their order and the Version dropdown to pin a module version. When you chain modules, check that each output matches the next input: an image saved as an artifact is not necessarily the value passed to the next step. Select Next.

4. Set the parameters

Under Function parameters, enter the inputs each module needs. NDVI Index takes the one-based band indices for the near-infrared and red bands.

Select Run Analysis. The data tree shows Pipeline run scheduled. with a View jobs button.

Monitor and inspect the result

Open Jobs. The new run appears with the type Pipeline run and its ID, such as #42; the same ID identifies the run in the SDKs, the API, and the Agent. Active lists runs that are pending, queued, or running.

Select the row, or View details and log in its ⋮ menu, to follow the Processing log. The request is queued for a worker, so allow the run to finish before interpreting its outputs.

If it fails, read the error in the log. Check that the dataset contains files matching the module's supported inputs, that the band indices fit the sensor, and that the module's dependencies installed. Fix the cause, then use Rerun pipeline from the ⋮ menu.

When the run reads Completed, select View artifacts. Tabular results are saved as an Excel workbook; Viewing and Exporting Results describes every output.

Run from code

The Python and R clients and the CLI submit the same pipeline definition:

phenoworks pipeline-runs run --dataset-id 42 --file pipeline.json --wait

In Python, pass data_products=["orthomosaic"] or modalities=["multispectral"] to client.run_pipeline(...) to send blocks only part of each plot.

For large datasets of independent plots, add "batch_size" to the run's parameters; see batched pipeline execution.

Try image processing next

For RGB data, the bundled histogram_equalization module adjusts contrast and returns a table with one row per processed image. Its source is blocks/image_analysis/histogram_equalization.py. Review a few images first: whether histogram equalization improves a particular dataset needs visual assessment.

Continue with Viewing and Exporting Results, or run an analysis conversationally with the Agent.