Zoho DataPrep helps you clean and transform your data based on prompts to Ask Zia in Natural Language. Using this feature, you can enter the required prompt, and the rest of the output will be auto-generated for you by Zia. The Ask Zia feature is powered by ChatGPT API, a generative artificial intelligence developed by OpenAI. To enable this feature in your DataPrep organization, go to the OpenAI Integration tab in the Settings page. Click here to know more about enabling OpenAI integrated features.

Data Privacy Notice : Zoho DataPrep utilises OpenAI's ChatGPT API, an external service based in the USA, subject to OpenAI's Terms of Use . Please be aware that the information sent to OpenAI depends on the access provided by you, the user. The data shared with ChatGPT is based on the features you enable. Details such as column names and/or selected cell data will be shared based on each feature. You can choose to enable or disable each feature based on the level of data you want to share. Therefore, exercise caution while using this service and ensure that the data you provide does not contain confidential or personal information.

Go to OpenAI Integration in the Settings page to learn more about configuring this integration.

 

Notes :

 

1. Ask Zia - Chat DataPrep is optimised for GPT-4 (not compatible with older versions).

2. The generative response is coming in from ChatGPT's training data. Accuracy, consistency, and relevance of the response are subjective to the type of prompts and the data available in its training set.

3. ChatGPT doesn't respond to prompts about personal information.

Prepare data using Ask Zia  

 1. Click the Ask Zia icon from the top right corner of the Studio page.

2. Provide a hint to clean and transform your data in Natural language in the Command Prompt section. For example, Convert all date columns to the 'MM/DD/YYYY' format. You can also insert one of the prompts from the sample suggestions.

3. Click Enter or icon and you will see a preview of the data generated by ChatGPT.

Info: You can hover over the preview data to view the original value before transformation.

Info: You can click on the column name displayed at the top left corner to navigate to the modified column.

4. Click Apply Transform to apply the transform to the entire data.

Info:

1. You can use the clone icon   to copy the interpretation or command, tweak it, or add new conditions to enhance your prompt.

2. When you close the Ask Zia window, the chat entries, including all prompts and interpretations will be cleared off the memory.

3. Each prompt is treated independently. Unlike ChatGPT, where context from previous prompt is used, here with Ask Zia, you will have to provide prompts from scratch each time. Conversation threads aren't maintained with Ask Zia - Chat DataPrep feature.

5. Once the transform gets applied to the data, it is added as a new rule in the ruleset.

On clicking the Edit option, you can modify the rule.

Note: Please note that the transformation was already applied. When you make edits to the prompt, it will modify the rule and affect the data again.

Note:

1. Only the prompt you type in is shared with ChatGPT.

2. For better results, please include an example with the prompt you provide.

Here are some sample prompts for you to try out  

1. I would like to fill the empty cells in all the columns with "Unknown". Below is a snapshot of the prompt and the preview of the output.

Prompt: Fill empty cells in all columns with "Unknown".

2. I want to clean the phone numbers in my dataset by removing the delimiters and standardising them. Below is a snapshot of the prompt and the preview of the output.

Prompt: Standardise the phone numbers by removing spaces and special characters in the 'Mobile' column

3. I want to remove the prefix 'mailto:' from the Email ID column. Below is a snapshot of the prompt and the preview of the output.

Prompt: Remove the prefix 'mailto:' from all the values in the 'Email' column.

4. I want to remove the duplicates from the data with the 'Email' column as reference, and I would like to remove all emails with public domains. Below is a snapshot of the prompt and the preview of the output.

Prompt: Remove duplicates from the data with the 'Email' column as reference and remove all the emails with public domains.

5. I need to merge the first name and last name into full name. Below is a snapshot of the prompt and the preview of the output.

Prompt: Merge 'First name' and 'Last name' to create a new column 'Full Name'.

6. I want to count the total leads for each state, add this count as a new column, and then sort the data by this count in descending order. Below is a snapshot of the prompt and the preview of the output.

Prompt: Give the total lead count based on the 'State' column and add the Total lead count in a new column retaining all other columns, sort by descending order.