Olivia Lounsbury’s Post

This Vanderbilt class was the hardest ChatGPT course I have run across. I passed every test the first go-around, but Weeks 4 and 6 were brutal. Had to really work and think about what I was doing. Weeks 1, 2, 3, and 5 I was already there. I will be taking the ChatGPT Advanced Data Analysis course next. Dr. White runs a good online course. Word of warning - the tests get harder and harder! ChatGPT is incredible in terms of what you can do. A real game-changer for Proposals and Business Development (BD).

Project: Data cleanup in Excel, sourcing terrible data sources from several products and collating to streamline reporting for an annual project. Names aren't consistent, SMSF handling isn't consistent, one replaces '&' with "+', etc. Just a right mess. I can't understate how bad it is. April 2022: Did some Excelfu to help streamline the process, but large parts were manual. April 2023: Was able to use ChatGPT go way further than before, cleaning up data in a way that would've completely broken me without it. Small parts needed manual review, but were so few it was easy to do manually. April 2024: ChatGPT is eating up my (our? it's own?) formulas from last year, dissecting, and spitting them back out better than ever. Being a daily user, I'd been getting a bit disenchanted from my subjective experience over time, but to see such an objective change solving the same problem so much better over twelve months is very cool.

ChatGPT 4 with Code Interpreter is powerful. I fed it a raw spreadsheet containing our LinkedIn Post analytics for the last 12 months. Without any data manipulation, it could quickly understand the data and began to perform analysis using basic prompts. It did a few strange things with the import, like ignoring a couple of columns and renaming one from 'reposts' to 'shares.' Using prompts, I began asking it questions about our visits and what drove engagement. As you can see from the analysis, I couldn't find anything actionable.

ChatGPT vs Bard – Summary Statistics Requested min, max & avg results of water temperature values included in pasted tabular data with multiple columns from Excel. 🔹 ChatGPT provided correct min & max with a slight error in the avg. When prompted, ChatGPT acknowledged the rounding error in the avg result. 🔹 Bard provided incorrect min, max and avg values, indicating an inability to read the data as pasted from Excel. When provided the data in .csv format, Bard was able to provide correct statistics, except for the avg. When prompted, Bard also acknowledged a rounding error in the avg. 💡 - Both systems had difficulty with correctly rounding the average result. - ChatGPT appears more capable with interpreting different data formats. - Bard required a more simple/standardized format to interpret the data.