Contents
How do you select a predictive model?
What factors should I consider when choosing a predictive model technique?
- How does your target variable look like?
- Is computational performance an issue?
- Does my dataset fit into memory?
- Is my data linearly separable?
- Finding a good bias variance threshold.
What are the characteristics of a predictive model?
Predictive analytics uses predictors or known features to create predictive models that will be used in obtaining an output. A predictive model is able to learn how different points of data connect with each other. Two of the most widely used predictive modeling techniques are regression and neural networks.
How are predictors added to a forecast model?
Predictors are added one at a time, and the one that most improves the measure of predictive accuracy is retained in the model. The procedure is repeated until no further improvement can be achieved. Alternatively for either the backward or forward direction, a starting model can be one that includes a subset of potential predictors.
When to choose the correct predictive modeling technique?
Selecting the correct predictive modeling technique at the start of your project can save a lot of time. Choosing the incorrect modeling technique can result in inaccurate predictions and residual plots that experience non-constant variance and/or mean.
Which is the best method to select predictors?
Maximising ¯R2 R ¯ 2 works quite well as a method of selecting predictors, although it does tend to err on the side of selecting too many predictors. Time series cross-validation was introduced in Section 3.4 as a general tool for determining the predictive ability of a model.
How to select predictors in a multiple regression model?
In most of the examples in this book, we use the AICc value to select the forecasting model. In the multiple regression example for forecasting US consumption we considered four predictors. With four predictors, there are 24 = 16 2 4 = 16 possible models.