@@ -0,0 +1,15 @@
+# -*- coding: utf-8 -*-
+"""
+This graph is limited to 3 levels. 
+With 2 chosen variables CLUSTER2, CLUSTER1, and target as SCH_YR this graph can be interpret as:
+Overall in test set, there are 2544 positive, 1358 negative.
+For CLUSTER1 less than 294, there are 2324 positive, 1350 negative.
+For CLUSTER1 less than 103.5, there are 2 positive, 24 negative. 
+For CLUSTER1 more than 103.5 and less than 294, there are 2322 positive, 1326 negative. 
+...
+in third level, I end up with 8 node, where there is some node with high probability of positive, and with big enough size (>30) such as:
+leaf 3: CLUSTER1 less than 107 and larger than 103.5
+leaf 5: CLUSTER2 less than 449, CLUSTER1 less than 420.5
+leaf 7: CLUSTER2 larger than 449, CLUSTER1 less than 422
+
+"""
@@ -0,0 +1,95 @@
+# -*- coding: utf-8 -*-
+"""
+Spyder Editor
+
+This is my submission to Machine Learning Course on Coursera. 
+This containts heavily sample code provided in the course. 
+I changed the dataset, test ratio 0.2, and max_depth = 3
+I also have to use write_pdf instead of create_png to create output
+Input file used: 
+
+"""
+
+# -*- coding: utf-8 -*-
+
+from pandas import Series, DataFrame
+import pandas as pd
+import numpy as np
+import os
+import matplotlib.pylab as plt
+from sklearn.cross_validation import train_test_split
+from sklearn.tree import DecisionTreeClassifier
+from sklearn.metrics import classification_report
+import sklearn.metrics
+
+os.chdir("C:\Machine_learning")
+
+"""
+Data Engineering and Analysis
+"""
+#Load the dataset
+
+AH_data = pd.read_csv("addhealth_pds.csv")
+#read data
+data_clean = AH_data.dropna()
+#drop all NA in dataset
+
+#data_clean.dtypes
+
+# data_clean.describe()
+#examine and summary data
+
+"""
+Modeling and Prediction
+"""
+#Split into training and testing sets
+
+#predictors = data_clean[['BIO_SEX','CORE1','CORE2','HIEDBLK',
+#'age','H1GI1M','H1GI1Y','H1GI2','H1GI6E','H1GI7F',
+#'H1GI14','SMP07','CLUSTER1','CLUSTER2','H2GI1M']]
+
+predictors = data_clean[['CLUSTER2','CLUSTER1']]
+
+
+#predictors.dtypes
+
+#define many data fields as predictors,
+
+targets = data_clean.SCH_YR
+# and one for target (dependent variables)
+
+pred_train, pred_test, tar_train, tar_test  =   train_test_split(predictors, 
+                                                targets, test_size=.2)
+# with the order of train-test and pred-target, this train_test_spit() 
+# function cut predictior set and target set into two: train subset and test subset. 
+# test_size determine the relative size of test subset
+
+pred_train.shape
+pred_test.shape
+tar_train.shape
+tar_test.shape
+#check the shape (what it means??) of four subset created 
+
+#Build model on training data
+classifier=DecisionTreeClassifier()
+classifier=classifier.fit(pred_train,tar_train)
+ #make predictions
+predictions=classifier.predict(pred_test)
+
+sklearn.metrics.confusion_matrix(tar_test,predictions)
+sklearn.metrics.accuracy_score(tar_test, predictions)
+
+#Displaying the decision tree
+from sklearn import tree
+#from StringIO import StringIO
+from io import StringIO
+#from StringIO import StringIO 
+from IPython.display import Image
+out = StringIO()
+tree.export_graphviz(classifier, out_file=out,max_depth=3)
+
+
+
+import pydotplus
+graph=pydotplus.graph_from_dot_data(out.getvalue())
+Image(graph.write_pdf('love2.pdf'))