Is it possible to use the ols function with the context by clause in DolphinDB?

Question

When I use the ols function with the context by clause as follows, an error occurs: update factor set cor = moving(calOls{n}, (close, arc, vrc, src, krc), win) context by underlying => calOls: r = ols(nextRet, a) => The input matrix is singular.

update factor set o = ols(nextRet, (arc, vrc, src, krc))[0] context by underlying

The following is my script:

defg calOls(n, close, arc, vrc, src, krc)
{
    tb = table(close as close, arc as arc, vrc as vrc, src as src, krc as krc)
    update tb set nextRet = log(close.move(-n)/close)
    b =  matrix(tb[["arc", "vrc", "src", "krc"]])
    try
    {
        r = ols(tb["nextRet"],b)
        return r[0] + r[1] * arc[size(arc)-1] + r[2] * vrc[size(vrc)-1] + r[3] * src[size(src)-1] + r[4] * krc[size(krc)-1]
    }
    catch(ex)
    {
        print(b)
        print("=========")
        return 0
    }
}

Answer 1

The error occurs because the input matrix is non-singular. To address this, you can add a logic to check whether the input matrix “b” is non-singular, as shown in the following script:

if(sum(0!=svd(b, computeUV=false)) == 0){
        return 0    
}