Applying Rigorous Machine Learning Methods in Business Strategy
summary
Data is profiled with pandas profiling, a high‑missing smoker column is dropped, numeric gaps get group‑averaged imputation, and categorical fields get one‑hot encoding. Models run in scikit‑learn, RapidMiner and h2o AutoML (SVM with RBF kernel, KNN, random forest, gradient boosting); PCA reduces dimensions and boosts accuracy above 90% while h2o’s leader GBT hits AUC ≈99.3%. A final classifier is exported as a pickled model and used to return churn probabilities (example prediction ≈31% leave for a sample employee).
AI-written summary · about
chapters (16)
- 0:00Rigorous ML in business strategy
- 3:33Data distributions and correlations
- 7:44Missing data and feature engineering
- 12:13Domain expertise and visualization
- 24:58Exploratory analysis and bivariate plots
- 28:18Automated feature engineering in RapidMiner
- 35:44Building predictive models from scratch
- 45:27SVMs and hyperparameter tuning
- 51:03Random forests and model validation
- 57:15PCA for dimensionality reduction
- 1:01:08Operationalizing models with MOJO/pickle
- 1:10:47Ethics and bias in HR analytics
- 1:15:26H2O AutoML vs RapidMiner
- 1:22:34Deep learning and gradient boosting
- 1:28:06ML applications across industries
- 1:31:43ML responsibility and societal impact
related talks
22mPredict the future (or don't) with machine learning in applications
30mScalable Automatic Machine Learning with H2O
56mPredictive Analytics for Practitioners
23mCustom-made games with machine learning and big data
02h 00mData Science Driven Business Insights with Python and PySp
30mMachine Learning Techniques for Class Imbalances & Adversaries
22mReal-world applications of machine learning: Beyond the hype
44mScalable Automatic Machine Learning with H2O