Machine Learning for a Rescue

Explore machine learning concepts and applications in a 53-minute conference talk from code::dive 2016. Dive into real-world problem-solving using ML techniques, starting with a client issue from the RESCUE Source Ministry. Progress through various approaches, from basic if-statements to sophisticated machine learning workflows. Examine classification, regression, clustering, and other ML paradigms. Learn about supervised, unsupervised, and reinforcement learning. Analyze a developers dataset using regression, K-means clustering, and neural networks. Discover how to automate the classification of junior and senior developers. Emphasize the importance of focusing on ideas rather than specific tools, and understand machine learning as a comprehensive process involving problem definition, data analysis, algorithm selection, and model validation.

Intro
RESCUE
Source Ministry
CLIENT PROBLEM
M BACKLINKS CLASSIFY THEM
1ST APPROACH IF-OLOGY UGLY CODE FOR POC
ND APPROACH NAIVE MACHINE LEARNING
DOING WITHOUT KNOWING ISA. RECIPE FOR A FAILURE
RD APPROACH, FINAL DATA ORIENTED MACHINE LEARNING WORKFLOW
CLASSIFICATION REGRESSION CLUSTERING DIMENSIONALITY REDUCTION ASSOCIATION RULES
SUPERVISED LEARNING UNSUPERVISED LEARNING REINFORCEMENT LEARNING
OUR PROBLEM
DEVELOPERS DATASET
REGRESSION PREDICTING VALUES
CLUSTERING K-MEANS
1936, RONALD FISHER IRIS DATASET
RESULTS STABILITY
CLASSIFICATION FAST ARTIFICIAL NEURAL NETWORK
HOW TO CLASSIFY OUR DATASET AUTOMATED WAY TO FIND JUNIOR/SENIOR DEVELOPER?
TECHNOLOGY
FOCUS ON IDEAS NOT TOOLS
ML IS NOT A SINGLE RUN
IT'S A PROCESS
DEFINE A PROBLEM ANALYZE YOUR DATA UNDERSTAND YOUR DATA PREPARE DATA FOR ML SELECT & RUN ALGO(S) TUNE ALGO(S) PARAMETERS SELECT FINAL MODEL VALIDATE FINAL MODEL