Citizen data science
summary
Claim: non-coders can build and publish a usable regression model with low effort using Azure Machine Learning Designer. Demo uses a CSV housing dataset, a provisioned compute cluster, column selection, 0.8/0.2 train/test split, linear regression, train/score/evaluate modules and shows R² around 0.61; example predictions (233k vs actual 260k) are imperfect. Model can be published as a REST endpoint for apps, though live deployment sometimes hiccups; speaker shows concrete steps, numbers and the model workflow rather than high-level theory.
AI-written summary · about
chapters (23)
- 0:00Start
- 3:24The Growth Mindset
- 3:56How Much Data Do You Think Is Generated every Year
- 7:42What Do You Believe Is Data Science
- 11:47What Data Scientists Are
- 14:24Professional Requirement
- 23:06Deploying an Azure Machine Learning Instance
- 23:26Storage Accounts
- 24:09Azure Portal
- 24:31Deploy an Azure Machine Learning Service
- 26:06Azure Machine Learning
- 30:33Uploading a Data Set
- 33:52Schema Selection
- 37:12Example Data Sets
- 39:11Checking for Outliers
- 39:40Clean Missing Data
- 42:23Split Data
- 43:54Train Model
- 46:29Linear Regression
- 47:09Linear Regression Model
- 52:12The Coefficient of Determination
- 55:05Correlation Matrix
- 57:38Correlation Is Not Causation
related talks
01h 05mAzure Machine Learning for the Developer
59mPredicting the future as a service with Azure ML and R
48mMachine Learning in Azure Databricks
01h 03mMachine Learning for Muggles
41mLeveraging Automated Machine Learning to enable anyone to develop Machine Learning solutions
45mMachine Learning for .NET developers
53mImplementing AI using Azure Machine Learning Services
59mA developers guide to Machine Learning