Partial Dependence plot¶

ExpyBox.pdplot()¶

Create dialog for partial dependence plot

Method parameters¶

Feature to plot
Which feature should be plotted
Number of grid points
Number of grid points to calculate partial dependence value in (for numeric feature)
Grid type
Type of grid points for numeric feature (percentile or equal)
Custom grid range
- Custom range minimum/maximum
  Percentile (when grid_type = percentile) or value (when grid_type = equal) lower/upper bound of range to investigate (for numeric feature)
  
  (Enabled only when Custom grid range is checked and Variable with grid points is None)
- Variable with grid points
  Name of variable (or None) from kernel_globals with customized list of grid points (list of integers) for numeric feature
Center the plot
Whether the plot should be centered (starting at [0,0])
Plot data points distribution
Whether to display data points distribution below the plot
Show precentile buckets
Whether to show percentile buckets below the plot
Plot lines - ICE plot
Whether to plot also lines for single instances (which makes the plot an ICE plot)
- Lines to plot
  How many lines to plot, can be a integer or a float
  
  integer values higher than 1 are interpreted as absolute amount
  
  floats are interpreted as fraction (e.g. 0.5 means half of all possible lines)
- Cluster lines
  Whether to cluster the lines
  
  Number of cluster centers - Number of cluster centers for lines
  
  Cluster method - Method to use for clustering of lines (KMeans or MiniBatchKMeans)