9.6 AN APPLICATION OF A NEURAL NETWORK MODEL
We next turn to an example of a neural network model using a subset of the Framingham Heart Study data.2The data set, Framingham_training, contains information on three variables for 7953 patients. Sex is a binary predictor with 1 = Male and 2 = Female. Age is a continuous predictor. The target variable is Death, with values 0 = survival and 1 = death.
Clues to the relationship between the predictors and the target are obtained through exploratory data analysis, namely through Figures 9.5 and 9.6, and Tables 9.2 and 9.3. The histograms in Figures 9.5 and 9.6 show that, as Age increases, the proportion of Death increases. Tables 9.2 and 9.3 show that a larger proportion of males died, compared to females. Thus, these interrelationships should be reflected in our neural network model results.
![Histogram from R of Age with death overlay displaying stacked bars with shades representing death 0 (light) and 1 (dark).](https://static.packt-cdn.com/products/9781119526810/graphics/images/c09f005.gif)
Figure 9.5 Histogram from R of Age, with Death overlay.
![Normalized histogram from R of Age, with Death overlay depicting staked bars in discrete shades representing 0 (light) and 1 (dark).](https://static.packt-cdn.com/products/9781119526810/graphics/images/c09f006.gif)
Figure 9.6 Normalized histogram from R of Age, with Death overlay.
TABLE 9.2 Contingency...