Team Name:
InsightPilot
Team Members:
InsightPilot
InsightPilot An AI driven chatbot interface that allows dataset operations to be performed using natural language, helping users gain no-code insights from datasets.
It uses a Q&A Sentence Transformer model from Hugging Face (multi-qa-MiniLM-L6-cos-v1) to interpret and map natural language queries such as "give me insights on sick leave usage" to predefined functions that interact with the dataset, which will pull up sick leave usage as a response. It provides transparency, as it shows the user what function it is picking, and how confident it is in its decision as a decimal score from 0 to 1.
We initially wanted to use the HR Dataset, but found it hard to gain insights from, so we switched to the Employee Leave Dataset, as we found it was better annotated and easier to draw meaning from. From this dataset, we devised what insights a HR employee may want from this data, such as monitoring sick leave usage, visualizing monthly leave data, and check who isn't taking enough leave (overworking) or who is close to using up their entitlement.
Description of Use Use in demonstrating our chat bot's query capacity
Go to Challenge | 18 teams have entered this challenge.
