This is an artificial data set with dependencies between the attribute
values. The cases are generated using the following method:

*X*_{1} : uniformly distributed over
[-5,5]*X*_{2} : uniformly distributed over
[-15,-10]*X*_{3} : IF (*X*_{1} > 0) THEN
*X*_{3} = green

ELSE *X*_{3} = red
with probability 0.4 and *X*_{4}=brown with prob.
0.6*X*_{4} : IF (*X*_{3}=green) THEN
*X*_{4}=*X*_{1}+2*X*_{2}

ELSE *X*_{4}=*X*_{1}/2 with prob. 0.3, and
*X*_{4}=*X*_{2}/2 with prob.
0.7*X*_{5} : uniformly distributed over
[-1,1]*X*_{6 }:
*X*_{6}=*X*_{4}×[epsilon], where [epsilon] is
uniformly distribute over [0,5]*X*_{7} :
*X*_{7}=yes with prob. 0.3 and *X*_{7}=no with prob.
0.7*X*_{8} : IF (*X*_{5} < 0.5) THEN
*X*_{8} = normal ELSE *X*_{8 }=
large*X*_{9} : uniformly distributed over
[100,500]*X*_{10} : uniformly distributed integer over the
interval [1000,1200]

Obtain the value of the target variable *Y* using the rules:

IF
(*X*_{2} > 2 ) THEN *Y* = 35 - 0.5
*X*_{4}

ELSE IF (-2 <= *X*_{4} <= 2) THEN
*Y* = 10 - 2 *X*_{1}

ELSE IF (*X*_{7} = yes)
THEN *Y* = 3 -*X*_{1}/*X*_{4}

ELSE IF
(*X*_{8} = normal) THEN *Y* = *X*_{6} +
*X*_{1}

ELSE *Y* = *X*_{1}/2

- Source: Contact me (ltorgo@liacc.up.pt).
- Characteristics: 40768 cases, 10 attributes (3 nominal, 7 continuous).
- Download : mv.tar.gz(132146 bytes)