Deploying machine learning models for forensic anthropology
summary
Models predict skeletal sex from cranial and long-bone measurements with probability scores, trained on two public osteoarchaeology datasets totaling thousands of individuals with global coverage. Implementation uses pandas, scikit-learn and XGBoost, models exported with pickle, wrapped into a SexEst web app built in Streamlit and containerized with Dockerfile and requirements.txt. The docker image builds reproducibly, the container runs Streamlit on port 8501, and the deployed site and local runs both operate successfully.
AI-written summary · about
chapters (10)
- 0:00Intro
- 0:27What is machine learning?
- 1:47The problem
- 4:18How to use the models
- 6:36Why Streamlit
- 8:45What is Docker?
- 12:03Docker images - Docker Hub
- 13:10Docker images - Dockerfile
- 16:19Docker images - requirements.txt
- 17:00Docker images - final result
related talks
41musing-django-docker-and-scikit-learn-to-bootstrap-your-machine-learning-project/
54mRipped Machine Learning with Distributed Computing
46mProduction Model Deployment
35mGetting Machine Learning Models Ready For Production Using Python
01h 02mDocker for Data Science: Reproducibility and Deployment
32mMachine Learning Infrastructure at Stripe: Bridging from Python JVM
01h 45mBuilding a Model Prediction Server
50mBuilding Intelligent Mobile Apps using Machine Learning