Completed Sheetseeker Assignment

[AmanKishore]

Finance Sheet Analysis with GPT-4-turbo

Summary

This PR leverages GPT-4-turbo to determine what information the user is querying for and returns a new xlsx sheet with the relevant data highlighted.
I created the SheetAnalyzer class to call GPT-4-turbo, which returns a list of financial terms relevant to a given query. These terms are then used to highlight corresponding values in the spreadsheet.

Query Processing with GPT-4-turbo

The get_relevant_rows function in utils.py is the cornerstone of our query processing. It constructs a conversation with the GPT-4-turbo model, providing it with a list of financial terms from the first column of the spreadsheet and the user's query. The conversation is formatted to prompt GPT-4-turbo to respond with a list of relevant financial terms.

• System Message: Sets the context for GPT-4-turbo, describing its role as a financial assistant.
• User Message: Contains the actual instruction, formatted using the INSTRUCTION template from constants.py, which includes the list of financial terms and the user's query.

The GPT-4-turbo model's response is expected to be a list of strings. If the response is not in the correct format, a retry mechanism is triggered, which attempts to correct the format using JSON mode.

Highlighting Relevant Rows

Once the relevant rows are identified, the SheetAnalyzer class's find_and_highlight method iterates over each sheet's DataFrame. It checks if the first column's value of each row is in the set of relevant rows returned by GPT-4-turbo. If a match is found, it highlights all numeric values in that row using the specified PatternFill.

Saving Highlighted Sheets

After highlighting, the workbook is saved in a new directory named highlighted_data. The file name is constructed by sanitizing the query string to ensure it is a valid file name.

Installation and Usage Guide for SheetAnalyzer

Prerequisites

Before proceeding, ensure that Python 3 is installed on your system by running the following command in your terminal:
python3 --version

Installation Steps

Step 1: Clone the Repository

If applicable, clone the repository to your local machine using the following commands:
git clone https://github.com/your-username/sheetseeker.git
cd sheetseeker

Step 2: Install Dependencies

Install the necessary Python packages listed in requirements.txt:
pip3 install -r requirements.txt

Step 3: Set OpenAI API Key

You need to set the OPENAI_API_KEY environment variable with your OpenAI API key. Replace your_api_key_here with your actual API key:
export OPENAI_API_KEY='your_api_key_here'

Running the Script

Once the setup is complete, you can run the solution.py script from the root directory of the project:
python3 src/solution.py

The script will process the predefined queries and highlight the relevant rows in the Excel spreadsheet. The highlighted spreadsheets will be saved in the highlighted_data directory within the project.

Conclusion

By integrating GPT-4-turbo's powerful natural language processing capabilities, SheetAnalyzer enables the user to highlight relevant information from financial documents for any query.