RUN_AB_TEST_ANALYSIS
Performs statistical A/B testing (T-test) on a dataset to determine if there is a statistically significant difference between two variants/groups.
Parameters
| Name | Type | Required | Description |
|---|---|---|---|
input_path | string | yes | Absolute path to the CSV/Excel dataset. |
variant_column | string | yes | The categorical column name containing the variant labels (e.g., 'Group', 'Variant'). |
metric_column | string | yes | The numerical column name containing the metric to test (e.g., 'Conversion_Rate', 'Revenue'). |
control_variant | string | yes | The exact name of the control group in the variant_column (e.g., 'A', 'Control'). |
treatment_variant | string | yes | The exact name of the treatment group in the variant_column (e.g., 'B', 'Treatment'). |
How to use it
You normally trigger this by describing what you want in chat — the agent selects RUN_AB_TEST_ANALYSIS automatically. For example:
Try saying
“use analyst tools to run …”
In a workflow
As a step in a multi-step workflow DAG:
json
{
"id": "s1",
"agent": "analyst_tools",
"action": "RUN_AB_TEST_ANALYSIS",
"args": {
"input_path": "/Users/me/Documents/file.txt",
"variant_column": "…",
"metric_column": "…",
"control_variant": "…",
"treatment_variant": "…"
},
"depends_on": [],
"outputs": []
}Direct call
For scripting, call it directly via POST /execute_tool. Every tool returns { success, message, data }.
bash
curl -X POST http://127.0.0.1:8000/execute_tool \
-H "Content-Type: application/json" \
-d '{"tool_name":"RUN_AB_TEST_ANALYSIS","args":{"input_path":"/Users/me/Documents/file.txt","variant_column":"…","metric_column":"…","control_variant":"…","treatment_variant":"…"}}'Part of the analyst_tools plugin. Browse the full Plugin & Tool Catalog or the relevant feature guide.