Using Artificial Intelligence for Finance Data Quality Management

Congratulations! You’ve made it to the last chapter of this book! In this chapter, we will be covering artificial intelligence (AI) and how various AI-powered solutions and techniques can address data integrity issues. Consider a scenario where a large dataset containing sales transactions has missing or incorrect details. Instead of fixing these data integrity issues manually, we will explore how we can use AI to automatically identify these discrepancies, suggest potential changes, and even perform necessary corrections. Exciting, right?

That said, these are the topics that we will cover:

• Introduction to AI
• Applications of AI in finance
• Detecting anomalies in financial transaction data
• Handling missing financial reporting data with AI
• Best practices when using AI for data integrity management

By the end of this chapter, you will have a better understanding of what AI-powered...

Technical requirements

The first two hands-on examples in this chapter utilize datasets we used in Chapter 5, Using Business Intelligence Tools to Fix Data Integrity Issues, and Chapter 6, Implementing Best Practices When Using Business Intelligence Tools. For a greater appreciation of our hands-on examples, it is recommended that you read these chapters. If not, it is still okay to jump right into it. The other requirements are as follows:

• A ChatGPT Plus account: Sign up and upgrade to the Plus version at https://chat.openai.com/auth/login
• Microsoft Excel (2016 version or later)

The datasets have been saved on GitHub. You can access them by going to https://github.com/PacktPublishing/Managing-Data-Integrity-for-Finance/tree/main/ch10.

Note

Once you are done with the hands-on examples in this chapter, feel free to downgrade your subscription and go back to the free version.

Introduction to AI

Remember the image shown in Figure 10.1 from Chapter 1, Recognizing the Importance of Data Integrity in Finance? You might be surprised that this image is AI-generated!

Figure 10.1 – The wolf hidden among the sheep

Here, the wolf hidden among the sheep symbolizes the subtle yet potentially catastrophic effects of data integrity issues. If you were to upload this image to an AI-powered solution such as ChatGPT, you would be amazed that it would be able to flag an anomaly in the uploaded image:

The image you've uploaded seems to depict a field teeming with sheep. However, upon closer inspection, there is a notable anomaly: amidst the sheep, there is a wolf. The wolf stands out because of its distinct appearance, which contrasts with the sheep. This inclusion of a wolf in a flock of sheep could be an intentional choice for a visual puzzle or to create a humorous effect, as wolves are predators of sheep and not typically seen...

Applications of AI in finance

The scope of AI for finance is expanding rapidly, thus enabling the automation of processes, assisting professionals in becoming more efficient at their jobs, as well as creating potential solutions to solve problems. That being said, here are some of the applications in finance in general:

• Outlier detection: AI helps in understanding the normal transactions to identify any anomalies.
• Credit scoring: ML models can process large datasets and use the patterns learned to predict creditworthiness and the ability to repay loans more accurately.
• Report generation: With AI, you can automate the creation of reports and contracts. One particular example is Alteryx AiDIN, which can be used to generate magic documents and help you create your first draft automatically so that it can be sent to key business partners.
• Missing data identification: As AI processes data efficiently, it can find missing and incomplete data in your datasets. This is...

Detecting anomalies in financial transaction data

In this example, we will work on the datasets we used in the Dealing with large financial datasets using data validation section in Chapter 5, Using Business Intelligence Tools to Fix Data Integrity Issues. We will test whether we can validate the data against another table containing the correct values. The steps are as follows:

1. Create a new chat using the Data Analysis GPT, as illustrated in Figure 10.7:
   

   Figure 10.7 – Using the Data Analysis GPT

   To access the Data Analysis GPT, click on Explore in the sidebar and select Data Analysis from the list of GPTs available. If you can’t see the GPTs, feel free to use the following link to access this: https://chat.openai.com/g/g-HMNcP6w7d-data-analysis.

Note

You may also search for chatgpt data analysis gpt via Google Search; you should find the GPT in the list of results. Alternatively, you may use GPT-4 as it has the Advanced...

Handling missing financial reporting data with AI

In this section, we will use ChatGPT to help us understand what our sample Excel files contain and check for any outliers, missing values, and data inconsistencies. We will be using datasets created specifically for this section to get more appreciation for this tool’s capabilities. The steps are as follows:

1. Create a new chat using the Data Analysis GPT, as shown in Figure 10.34. As an alternative, you may use the Advanced Data Analysis feature (using GPT-4), as shown in Figure 10.13.
   

   Figure 10.34 – Using the Data Analysis GPT

Note

If you can’t access the Data Analysis GPT from the Explore section in the sidebar, you may use this link: https://chat.openai.com/g/g-HMNcP6w7d-data-analysis.

2. Load the 2022 Transactions.xlsx, Products.xlsx, and Sales Price.xlsx files and type this prompt:

   Please examine the contents of these three files and provide information as to...

Best practices when using AI for data integrity management

Here are some of the best practices you can make use of when utilizing AI to improve the quality and integrity of your data:

• Clearly state your goals: Before you make use of an AI model, ensure that you are clear about what you want to accomplish. Knowing this will help guide you in your decisions in selecting the AI tool or developing the model to use for data cleansing, validation, governance, and compliance.
• Create clear and effective prompts: When using AI models, especially for generating content or answers to questions, it’s essential to prepare prompts that are concise and specific to the desired outcome. This gives rise to the importance of prompt engineering, which is a method of developing well-defined prompts to help ensure that the AI or language model understands the context and your requirements to produce relevant and accurate results.

Note

Creating well-structured prompts is key...

Summary

In this final chapter, we discussed how we can utilize AI for data quality management. At the start, we covered what AI is as well as its applications in finance. After that, we worked on hands-on examples to detect anomalies in financial transactions using a generative AI solution called ChatGPT. We gained an appreciation of how we can use this tool to produce similar results as we did in previous chapters. Then, we dived deeper into how we can use this tool to find missing data entries and values in our transactions and correct them. Lastly, we discussed the best practices to make the most out of AI-powered tools and solutions for data integrity.

Congratulations on finishing this book! I hope you found this to be an incredible journey considering we’ve covered different practical solutions to address various data integrity issues and challenges. If there are areas or topics that are unfamiliar to you, feel free to go back and review them again. At the same time...

Further reading

For more information on the topics we covered, feel free to check out the following resources:

• The economic potential of generative AI: The next productivity frontier:

  https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/the-economic-potential-of-generative-ai-the-next-productivity-frontier

• Gartner Experts Answer the Top Generative AI Questions for Your Enterprise:

  https://www.gartner.com/en/topics/generative-ai

• ChatGPT Code Interpreter: What It Is and How It Works:

  https://365datascience.com/trending/chatgpt-code-interpreter-what-it-is-and-how-it-works/

• Will ChatGPT Put Data Analysts Out of Work?:

  https://www.forbes.com/sites/bernardmarr/2023/02/07/will-chatgpt-put-data-analysts-out-of-work/?sh=76a468494030

• Embracing Generative AI: Opportunities and Risks for CFOs:

  https://www.forbes.com/sites/jackmccullough/2023/08/21/embracing-generative-ai-opportunities-and-risks-for-cfos/?sh=787b553057f7

• Introducing GPTs:

  https://openai...