I was asked to discuss the trade-offs of an in-development product that was similar to the technology I worked on at a previous employer.

Sections where you should add your own code and/or analysis are highlighted. Communicating your analysis with clear discussion and plots is very much appreciated. Feel free to add additional cells for any of your analysis and discussion. For this challenge you will be using raw 3-axis accelerometer (ACM) signals measured from a wearable device to build an automatic activity detection algorithm. We will be using the following two data files, which represent a single session from a single individual instructed to perform various activities in a laboratory setting. - **user1_accelerometer_data.parquet** - This pyarrow parquet file has raw accelerometer signals in x/y/z directions (in mill-g), with `time` (in seconds since epoch) continuously sampled at 50Hz. - **user1_activity_labels.csv** - This CSV file has the `time_start`, `time_end`, and `id` of each activity performed during the session.