addGpFunction
Note: This function is not supported by Community Edition. You can
get a trial of Shark from DolphinDB official website.
Syntax
addGpFunction(engine, func)
Arguments
engine is the engine object returned by
createGPLearnEngine.
func is a user-defined function. Currently it does not support complex
assignment,
if or for statement. Only
return statement can be used to return a combination of the
training functions (see Appendix for supported functions). For example:
def f(x, y){
return cos(x+y)
}Details
Add a user-defined training function to the GPLearn engine.
Examples
def f(x, y){
return cos(x+y)
}
addGpFunction(engine,f)Appendix
The following table lists available functions for building and evolving formulas.
| Function | Description |
|---|---|
| add(x,y) | Addition |
| sub(x,y) | Subtraction |
| mul(x,y) | Multiplication |
| div(x,y) | Division. If the absolute value of the divisor is less than 0.001, returns 1. |
| max(x,y) | Maximum value |
| min(x,y) | Minimum value |
| sqrt(x) | Square root of the absolute value |
| log(x) | Logarithm. If x is less than 0.001, returns 0. |
| neg(x) | Negation |
| reciprocal(x) | Reciprocal. If the absolute value of x is less than 0.001, returns 0. |
| abs(x) | Absolute value |
| sin(x) | Sine function |
| cos(x) | Cosine function |
| tan(x) | Tangent function |
| sig(x) | Sigmoid function |
| signum(x) | Returns the sign of x |
| mcovar(x, y, n) | Covariance of x and y within a sliding window of size n |
| mcorr(x, y, n) | Correlation of x and y within a sliding window of size n |
| mstd(x, n) | Sample standard deviation of x within a sliding window of size n |
| mmax(x, n) | Maximum value of x within a sliding window of size n |
| mmin(x, n) | Minimum value of x within a sliding window of size n |
| msum(x, n) | Sum of x within a sliding window of size n |
| mavg(x, n) | Average of x within a sliding window of size n |
| mprod(x, n) | Product of x within a sliding window of size n |
| mvar(x, n) | Sample variance of x within a sliding window of size n |
| mvarp(x, n) | Population variance of x within a sliding window of size n |
| mstdp(x, n) | Population standard deviation of x within a sliding window of size n |
| mimin(x, n) | Index of the minimum value of x within a sliding window of size n |
| mimax(x, n) | Index of the maximum value of x within a sliding window of size n |
| mbeta(x, y, n) | Least-squares estimate of the regression coefficient of x on y within a window of size n |
| mwsum(x, y, n) | Iner product of x and y within a sliding window of size n |
| mwavg(x, y, n) | Weighted average of x with y as weights within a sliding window of size n |
| mfirst(x,n) | First element of the window of size n |
| mlast(x,n) | Last element of the window of size n |
| mrank(x,asc, n) | Rank of x within the sliding window of size n |
| ratios(x) | Returns the value of x(n)/x(n−1)x(n) / x(n-1) |
| deltas(x) | Returns the value of x(n)−x(n−1)x(n) - x(n-1) |
Note: For moving window operators (such as
mmax(x, n)), the
parameter n, which represents the window size, is randomly selected from the
windowRange vector specified in the createGPLearnEngine
function. In the following example, the training process randomly selects a value
from [7,14,21] as the window
size.engine = createGPLearnEngine(source, predVec, windowRange=[7,14,21])