bokesyo / cart-de-novo Goto Github PK

Dear TA:

To run our program, you    「only」     need to run   「regMain.py」

0. It will automatically import all data from /inputData/partial_train.csv(80% of all trainng data)
                                                      remain_train.csv(20% of all trainng data)
                                                      test.csv

1. It will give a model using CART without pruning, and MSE

2. It will then use Weakest Link Pruning and cross validation to giving you a optimized tree, and MSE

3. It will then test the model over testing dataset and give you a MSE

4. It will then use M5' algorithm to improve the model, and give you a optimized tree and MSE

5. Then it will generate a function which can help you implement the model and output to /result/regressionTreeExecute.py

6. Then it will draw a graph of decision tree an output to /result/regressionTreeGraph.eps

7. All the tree object will be stored in /tmp/reg/forest/ (in .db form)



"""
If you want to know more about our program, you can refer to:

(The following modules will not run by themselves, you should call it in other program)

Compiler:
    compile.py

M5 Algorithm:
    M5.py

CSV reader:
    readData.py

Assessment Moduel:
    regAsses.py

Regression Tree Grow Module:
    regGrow.py

Tree Printer:
    treePrint.py

Tree Pruner:
    regPrune.py

"""