0A018G - Data science without a Ph.D. Using IBM SPSS Modeler (v18.1.1)

1: Introduction to data science and IBM SPSS Modeler

• Explain the stages in a data-science project, using the CRISP-DM methodology
• Create IBM SPSS Modeler streams
• Build and apply a machine learning model

2: Setting measurement levels

• Explain the concept of field measurement level
• Explain the consequences of incorrect measurement levels
• Modify a fields measurement level

3: Exploring the data

• Audit the data
• Check for invalid values
• Take action for invalid values
• Impute missing values
• Replace outliers and extremes

4: Using automated data preparation

• Automatically exclude low quality fields
• Automatically replace missing values
• Automatically replace outliers and extremes

5: Partitioning the data

• Explain the rationale for partitioning the data
• Partition the data into a training set and testing set

6: Selecting predictors

• Automatically select important predictors (features) to predict a target
• Explain the limitations of automatically selecting features

7: Using automated modeling

• Find the best model for categorical targets
• Find the best model for continuous targets
• Explain what an ensemble model is

8: Evaluating models

• Evaluate models for categorical targets
• Evaluate models for continuous targets

9: Deploying models

• List two ways to deploy models
• Export scored data