// //Regression Tree command // //This command makes regression tree of continuous target variable //by cathgory or numeric explanation variable. // //Syntax tree [target] by [explanations]/ target/continuous [options] ; //(caution) target variable must be continuous. //Options enforce/[variable] //[variable] is divided forcibly at first tree branch. test:0.xx //xx is test data share which is used for inspection of precision. terminal:nn //nn is minimum records at terminal node(default 100). [variable]:code //if variable type is numeric code you have to define code. [variable]:continuous //if numeric type you don't need to define explicitly. [variable]:moji //if character type you don't need to define explicitly. [variable]: // //Example 1 // //Prediction of stop distance at car speed. //This example is relationship with car speed and stop distance. get cars.csv@;
//Half of data is used as validation and other half is used as making data. tree dist by speed/ target/continuous terminal:3 test:20 ;
//show performance of varidation get score@ana;
plot scat speed predict;
Performance of varidation
//Test the predict value by liner regression. get car.csv@; reg dist by speed; get freq@ana; over line reg by speed;
Performance of varidation with over write of regression line.
//Example 2 // //regression tree analysis with cathgory explanation data. //Prediction of PetalL value by SepalL SepalW and cathgory 'Species variable. get iris.csv@;
tree PetalL by SepalL SepalW Species/ target/continuous terminal=3 test=50 ;
get score@ana;
Show table of predict and explanation variable.
plot scat actual predict;
Show comprison whith actual to predict value. This plots are alomost on rectangle line.