One of the most frustating aspects of this course is that it effectively requires learning an entirely new programming language spread across multiple platforms at the same time, which creates a steep and unnecessary cognitive load. The lectures are structured in a way that leaves the learner largely on their own unless they resort to copying and pasting the provided exercises which I did in earlier lession but —an approach that defeats the educational purpose of AI prompting learning. A major conceptual sticking point is the use of data frames within database notebooks. A data frame is introduced as an object that appears identical to a Python DataFrame, yet in practice it is often a database-managed or database-backed object with different execution semantics. The frame is somewhere in the mist of your imagination. The course never clearly or concisely explains this distinction—what runs in Python, what runs in the database engine, and how the object is represented across contexts—which leads to confusion at precisely the point where conceptual clarity is most needed. For example the Chat bot did not understand how to query the current envirnoment but after multiple onerous querys it finally gave the the correct code to find the most regions that ordered the ski products ………..grouped_df.size().reset_index(name=‘counts’) but bot gave multiple wrong code looking for a .cvs file not understanding that the df was in the loaded with the info?
Hi max_IO,
Good points.
The way I look at the course is that it provides an introduction on how to work with Streamlit and Snowflake. The learning curve may be somewhat steep at certain points, which may not make the course suitable for everyone at first run. But learners who actually start building with Streamlit and Snowflake can take the course again and again, acquiring more knowledge each time.
You are right that using the same dataframe terminology as pandas can be confusing, considering the objects appear to work differently behind the scenes. I guess I would use the term streamlit dataframe or snowflake dataframe, while reserving the term dataframe for a pandas dataframe.
Thanks for your feedback!
Thank you both for the great feedback and discussion. We do include a section that discusses the differences between a python and Snowflake dataframes, but I will take a look at how we can make that distinction more clear.
I would also love to hear more about where the learning curve is steep so we can improve those areas. Thank you!
Hi Summerla,
The speed of the second module is quite high. Certain code passages and scrolling go rather fast and learners may not have the time to digest what is shown unless they stop the video. Because solution notebooks are provided and no graded assignments are used, learners may not always go over the code carefully. And as chatbots do not always provide correct code output, debugging is often necessary. If the code shown in the course is not digested, it is hard for learners to debug code provided by the chatbot, and chatbots do not always succeed in resolving their own bugs. This is why I think taking the course a number of times is both useful and required if learners start building with Streamlit and Snowflake. It also strikes me as introductory, for instance because the course does not dive deeper into certain components of Streamlit, e.g. rerun and managing session_states between UIs. Some references to open source learning materials on such components may be interesting to learners.
Overall, I would say it is a very useful and effective, sometimes challenging, introductory course. Thanks for making it available!
You’re a great mentor!
Thanks so much for your feedback, this was helpful. I’m going to look into ways we can streamline the course because I do think it has a lot of very useful information about prototyping that I want to make sure is accessible for anyone that takes the course.