Data Science in Meatspace
summary
Practical playbook for real‑world sensing: use R and the Tidyverse + Python + RStudio Connect stack to ingest and iterate on physical measurements, quantify measurement system error and standardize data models, and always prototype with a GoPro or cheap sensor before buying industrial gear. Examples include plant phenotyping pipelines and a drum feeder vision system where video can only detect presence not count seeds, prompting a redesign. Emphasizes calibration, integration costs, and instrument selection that actually capture the signal needed.
AI-written summary · about
chapters (19)
- 0:00Intro
- 1:10What is Meatspace
- 1:55Bens background
- 5:49Challenges in Meatspace
- 6:38Data isnt the default
- 7:37The moth bug
- 8:20Genchi gembutsu
- 8:39R for Data Science
- 8:56Importing Data
- 9:25Pipeline
- 10:46Prototype
- 11:48Intuition
- 12:38Prototyping
- 12:49Instrumentation
- 14:30Example
- 15:29Calibration
- 17:00virtuous cycle
- 17:30thanks
- 17:55questions
related talks
39mDoing Data: The Critical Process of Data Preparation (...)
20mChampioning modern science workflows to benefit dairy farmers
14mWhy Food Tech Matters: Finding Value in Data
08mProcessing Robot Data at Scale with R and Kubernetes
36mFrom POC to Prod, a data science project laid bare
28mDataEngBytes 2023 - MEL-T1-11 - Emily Melhuish
39mHelp Me, Help You: Supporting Your Data
15mSave an ocean of time: streamline data wrangling with R