Opened 13 years ago

Closed 13 years ago

#359 closed request (fixed)

Smooth histogram

Reported by: Peter Owned by: Peter
Priority: major Milestone: yat 0.5
Component: statistics Version: trunk
Keywords: Cc:

Description

This algorithm can be viewed as an extension of a Histogram, and like a histogram it should be used to estimate a probability distribution. However, rather than binning data, a kernel function is used for smoother inclusion of data point.

The interface should resemble of the interface of Histogram:

  • Estimation points are defined in constructor telling in which points the probability distribution should be estimated.
  • One can add a data point (double) the the object.
  • A kernel function defines how to estimate distribution as d(y) = sum K(y,x_i)

Typically the Kernel function only depends on absolute value |y-x_i|. A normalization should be provided in same fashion as in Histogram (it need to be defined how this should behave for N -> inf)

The KernelFunctions? that exist in yat::regression could be used here, although this functionality does not belong in regression (rather in statistics, right?)

Histogram should be a special case of this algorithm, in which Rectangular Kernel Function is used and estimation points are equally distributed with a separation equal to twice the width of the Kernel Function.

Change History (2)

comment:1 Changed 13 years ago by Peter

Milestone: yat 0.x+yat 0.5
Status: newassigned

comment:2 Changed 13 years ago by Peter

Resolution: fixed
Status: assignedclosed
Note: See TracTickets for help on using tickets.