Data Cleaning (DC) Agent

The DC agent helps users to clean and organize their data within a workbook. This includes tasks like deduplication, normalization, data imputation, and filtering.

Key Features:

• Deduplication: Identifies and removes duplicate records based on selected columns.

• Normalization: Restructures data into multiple related tables to improve data organization and reduce redundancy.

• Data imputation: Fills in missing values (NAs) with user-specified values.

• Filtering: Allows users to filter data based on specific criteria.

• Conversational interaction: Users can interact with the agent using natural language commands (e.g., "help me dedup").

• Data analysis: Provides an overview of the data, including the number of rows and columns, unique values, and missing values.

• Downloadable output: Users can download the cleaned data.

• Reset functionality: Allows users to reset the agent's context and revert to the original data.

How to Use the Data Cleaning Agent

1. Select the DC Agent from the available AI agents.

2. Choose the Excel or CSV file that you want to clean.

3. Use natural language commands to perform data cleaning operations. For example:

   • "help me dedup" to start the deduplication process.

   • "fill 999 999 999" to fill missing phone numbers with that value.

   • "show me records with agent name Henry Madlin" to filter data.

   • "help me normalize" to normalize the data.

4. Review the cleaned data and download it using the provided button.

5. If you want to start over or clean the data differently, use the code command to revert to the original data.

• The DC agent is configurable by an administrator. If you cannot view the agent, contact your admin to enable it.

• The agent's context changes as you perform cleaning operations. Use the reset command to revert to the original data if needed.

• The original workbook remains intact, even after cleaning operations.

Example Usage

Tranformation Type	User Query Example
Deduplication	"help me dedup using agent name, policy number"
Data Imputation	"fill missing phone numbers with 999-999-9999"
Filtering	"show me records where agent name is John Doe"
Normalization	“help me normalize the data”