Deploying Chat Data Prep

Overview

This documentation provides a step-by-step guide on deploying Chat Data Prep within the Akkio platform. The purpose of this feature is to automate data cleaning and transformation processes, significantly reducing the time and effort required to prepare data for analysis and reporting. The deployment can be set up as a Web App or API, allowing for efficient handling of new datasets with the same schema.

Initial Setup

Load the initial dataset into a new project within Akkio. Apply the necessary data cleaning and transformation steps using Chat Data Prep. Refer to the Chat Data Prep page for more information.

Deployment

Once the data has been prepared and cleaned, you can deploy the transformations to automate the process for future datasets.

Navigate to the "Deployments" tab within your project. This tab will now be enabled.

Several endpoint options will be available. These endpoints include Web App, API, and several databases to deploy your dataset transformations to.

• If you are unsure on an endpoint, and are looking to upload files for automatic Data Preparation, select Web App.

Databases

Connect your account if you haven’t. Connection instructions can be found under each database’s Integration page. Next, select the dataset to apply transformations to.

API

An example curl command is provided. On the left hand side, a list of the dataset’s columns will be shown. Column select is only relevant to Model Deployment and has no effect on deployed data preparation.

Web App

Standalone Web App deployment for Chat Data Prep will create a unique url. You may return to this url at any time to upload files for automatic data preparation.

You’ll see that there are several options to tweak your Web App, including:

• Title - User-facing title for your web app.

• Description - Optional user-facing description for your web app.

• Select Fields - Select which fields to include. This feature is only relevant to Model Deployment and has no effect on deployed data preparation.

• Make Public - Allow anyone to access this app without requiring authentication—Yes or No.

• Allow Bulk Upload - Allow users to make bulk predictions by uploading datasets in CSV format. This must remain enabled to upload files for automatic data preparation.

Select "Deploy" in the top right corner to initiate the deployment.

A UI box will appear, noting that the deployment is live. Select, “Got to live app” in the top right corner to open the Web App.

Using the Web App

In the Web App, select "Upload Excel, CSV, Parquet, or JSON" to load a new raw data file. Ensure that the new file has the same schema as the original dataset used in the project.

Once the file is processed, you can download the cleaned data as a CSV file. This file can then be uploaded back into Akkio in a new project for further reporting and analysis.