How to upload data for AI Coding?

How to upload data for AI Coding?

To start a new coding project, go to the AI Coding Directory and click+ New AI Coding



Here is a quick demo using the inca AI Coding portal:
​


Step 1. Upload

You will be directed to the inca AI Coding studio to Upload Dataset. You can either drop the file or upload the data set. 

For the AI coding upload, users can upload a data file with one or multiple Verbatim columns, an optional Verbatim ID column, an optional Analysis Variable (can upload one or multiple columns), an optional Language column (when it's non-English). The dataset may be uploaded as Microsoft Excel (.xlsx), CSV (.csv), or SPSS (.sav) file.



From the file your have uploaded, select source for Verbatim text (your verbatim column), Verbatim unique IDs (your ID column), Analysis Variables (your Filter variable columns) and Language IDs (language IDs column). You may select multiple verbatim columns from a data file, the model will run AI coding separately for each verbatim column.


Verbatims - The verbatim column is simply the set of verbatims (spontaneous mentions from open-end questions) that you want to analyze. The AI coding supports up to 10,000 verbatims in each dataset or coding instance.

IDs - Verbatim IDs are identifiers uniquely associated with each verbatim, which will be retained during analysis and available in the export. Note that verbatim IDs are also optional. If IDs are not provided, then unique IDs will be automatically generated.

Analysis Variables (AI Coding Filters) - Users can also add Filter variables. Filters are useful analysis tool that allows you to filter the responses from the total sample and show only the coded data for a certain group/segment or data set that you have selected (i.e. By Gender, By Country, and Preferred Brand/Logo/Concept).

Language IDs - AI coding can natively handle in many languages. However, if you would like to use machine translation to translate verbatim to English before processing, simply include a language id variable/column in your dataset file.
The list of supported language IDs is available here. Note that Language IDs are also optional: if not provided, no machine translation will be performed.


Step 2.  Review Codeframe 
Users can use any of the following options for their Codeframe generation


1. Use an AI generated Codeframe 
2. Use an existing Codeframe
3. Upload a new Codeframe



Use an AI generated codeframe 

Use an AI generated codeframe and click Generate Codeframe.



Users can now review the AI-generated codeframe and optimize the output by enabling Brief AI to guide Codeframe generation and by adding a prompt. Adding a prompt or instruction to generate the new codeframe is similar to briefing a human coder. In the prompt, you may include their analysis objectives, what to include, and what to avoid in the codeframe.



Sample prompts:
- Generate a codeframe that reflects the feelings and emotions revealed in the verbatims, that is, how people felt about this rather than what they talked about this
- Based on the verbatims only, generate a codeframe on why this logo are preferred by consumers. What specific elements do they like. Is it the colour, font style, or size etc.

- Generate a codeframe of 2 net codes, functional benefits and emotional benefits and have 10 codes under each net accordingly


After adding the prompt, click Regenerate codeframe.

You can still modify the the generated codeframes before proceeding with the actual coding. You can add new codes, add new nets, merge codes on the codeframe.



Once you are satisfied with the codeframe, click Confirm Codeframe and Proceed with Coding.




Use an existing Codeframe

Users can use an Existing Codeframe saved from all the previously AI coding instance or analysis you have done on your account. 



Similarly, users can still modify the the existing codeframes before proceeding with the actual coding. You can add new codes, add new nets, merge codes on the codeframe. Once you are satisfied with the codeframe, click Confirm Codeframe and Proceed with Coding.




Upload a new Codeframe

Users can upload a new codeframe. A CodeFrame file is a CSV (.csv) file with a column labelled CodeLabel containing a list of code labels. It may also include the following optional columns: 
  • CodeID (unique identifier)
  • CodeSentiment ("Positive", "Negative", or "Neutral")
  • CodeNet (label of net category)
  • Example 1 (example verbatim)
  • Example 2 (example verbatim)
  • Example 3 (example verbatim)

The filename of the uploaded file will be used as the CodeFrame label. You may download codeframe_example.csv to see an example codeframe file.





Again, users can still modify the the Uploaded codeframes before proceeding with the actual coding. You can add new codes, add new nets, merge codes on the codeframe. Once you are satisfied with the codeframe, click Confirm Codeframe and Proceed with Coding.






Step 3.  Coding
Users will be notified via email once coding is completed. This could take a few moments to several minutes depending on the number of verbatim analyzed.  


Note: Once the codeframe is applied, you can no longer edit it for the current project. Nevertheless, when you have received the coded results, you can still review and optimize the coded results through the inca AI Coding Studio. Please refer to this article on how to modify coding here. 





    • Related Articles

    • How can I create a codeframe for AI coding?

      Human-in-the-Loop Codeframe Generation The human-in-the-loop (HITL) AI coding allows human researchers to prompt the AI to help shape the code frame generation. The HITL coding functionalities are available both on the inca dashboard and in the inca ...

    • How can I add Filter variables to AI Coding?

      When uploading your data in AI coding, users can also add Filter variables. Filters are useful analysis tool that allows you to filter the responses from the total sample and show only the coded data for a certain group/segment or data set that you ...

    • How can I create project "folders" to organize my AI Coding instance?

      Projects Feature Projects should be used as a way to organize AI coding instances visually on the directory. Each AI coding instance should only be assigned to one project. The simplest way to think of this feature is to think of the interaction that ...

    • How to Modify AI coding?

      Users can modify the results of AI coding whether they are (1) using the AI Coding app or (2) using the inca Dashboard 1. If you are using the AI Coding App as a separate tool, it means you have manually uploaded you data (verbatims file) in the AI ...

    • How can I save my codeframe and use it for another analysis?

      There are two ways in which users can save their codeframe and use for future cases or analysis: 1. Save the codeframe before proceeding with AI Coding 2. Save the codeframe after finalizing AI coding Here are the guidelines for each way: 1. Save the ...