Predictive analytics projects fail all the time. In this book, Jarlath Quinn draws on his experiences and those of his colleagues to explore precisely what drives success or failure in predictive analytics projects. When it works, why does it work and what can be done to manage the risk of failure? Download your free copy of this essential guide to the practicalities of predictive analytics in the real world.
- Defining predictive analytics
- ‘Tell us something we don’t know’
- Choosing a predictive analytics project
- Chapter 1 – Finding inspiration
-
- The CRISP-DM model
- Business understanding
- Making a plan
-
- Chapter 2 – The raw material
-
- Data understanding
-
- Chapter 3 – Shaping the data
-
- Data preparation
- Merging and appending files
- Aggregating data
- Transposing data
- Creating new fields
-
- Chapter 4 – The algorithm menagerie
-
- Predictive models
- Segmentation models
- Association Models
- Other model types
- The two cultures
- Statistical techniques
- Rule induction / decision trees
- Machine Learning
-
- Chapter 5 – Building a predictive model
- Chapter 6 – What does ‘good’ look like?
-
- Accuracy
- Interpretability
- Stability
- Coherence
- Simplicity
- Performance
- Visualising Model Performance
- Model performance metrics
- Overall accuracy
- Area under the curve
- Gini coefficient
- Lift
- Model validation
- Training / testing sample split
- Cross-validation
-
- Chapter 7 – Back in the real world
-
- Creating selections
- Testing deployment
-
- Chapter 8 – Beyond deployment
-
- Monitoring performance
- Automation
- Planning and deciding
- Last thoughts
-
