Question

You are to build trees with varying max tree depth for the dataset provided (use maximum tree depths 2-10). For each tree of a given maximum depth, record the accuracy, precision and recall. Plot each of these metrics as a line plot (tree depth on the x axis and % on the y axis). Below is what I have attempted.
In [1]: import pandas as pd import numpy as np from sklearn.model_selection import train_test_split from sklearn.neighbors import KNeighborsClassifier from sklearn.metrics import confusion_matrix In [2]: df = pd.read_csv('1.csv') df.head() Out [2] : sequential number x acceleration y acceleration z acceleration label 0 0.0 1502 2215 2153 1 1 1 1.0 1667 2047 2072 1957 2 2.0 1611 1906 1 3 3.0 1601 1939 1831 1 4 4.0 1643 1965 1879 1 In [3]: # Define the x columns (independent variables / features) and # y variable (response / target - what we want to predict) # first four columns X = df.iloc[:, 0:4] # Last column y = df.iloc[:, 4] # Generate the train and test sets X_train, X_test, y_train, y_test = train_test_split(x, y, random_state=0, test_size 0.2)

Answer 1

#!/usr/bin/env python
# coding: utf-8

# In[61]:


import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sklearn.metrics import confusion_matrix,precision_score,recall_score,accuracy_score


# In[62]:


from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier


# In[63]:


df=pd.read_csv("D:/advanced_ML_amit/Activity Recognition from Single Chest-Mounted Accelerometer/1.csv",header=None)


# In[64]:


print(df.head())


# In[65]:


# now lets separate dependent and indipendent Variable.(X and Y)
X=df.iloc[:,0:4] # first four column
print(X.head())


# In[66]:


Y=df.iloc[:,4] # last column(label)
print(Y.head())


# In[67]:


# now lets splits the dataset into train and test set.

x_train,x_test,y_train,y_test=train_test_split(X,Y,random_state=42,test_size=0.2)


# a loop is running from 2 to 10 which will be the  max_depth of the function DecisionTreeClassifier() and we store the scores of it .

# In[68]:


result={}
pre_list=[]
re_list=[]
acc_list=[]
for m_depth in range(2,11):
    clf_tree=DecisionTreeClassifier(random_state=0,max_depth=m_depth)
    clf_fit=clf_tree.fit(x_train,y_train)
    predicted=clf_fit.predict(x_test)
    pres_score=precision_score(y_test,predicted,average='macro')
    re_score=recall_score(y_test,predicted,average='macro')
    acc_score=accuracy_score(y_test,predicted)
    result[m_depth]=[pres_score,re_score,acc_score]
    pre_list.append(pres_score)
    re_list.append(re_score)
    acc_list.append(acc_score)


# In[69]:


print(result)


# In[70]:


x=np.arange(2,11)
plt.plot(x,pre_list)
plt.plot(x,re_list)
plt.plot(x,acc_list)
plt.legend(['precision', 'recall', 'accuracy'], loc='lower right')

plt.savefig('scores_line_plot.png')
plt.show()


# In the plot You can see higher the value of max_depth better the result but it almost saturated after max_depth=8.
# 
# so the question is can we increase the value of max_depth further for more better result?
# 
# it is very subjective to answer but larger the value of max_depth , the complexcity is higher.
# and after some value of max_depth scores will be saaturated so we need not to increase value of max_depth after that.
# 
# lets have a look the confusion matrix when max_depth=10.
# 

# In[71]:


print(confusion_matrix(y_test,predicted))


# only 13 points are wrongly predicted where 12 points are wrongly predicted label as 3 where these are actually label 5.
# and 1 points is wrongly predicted as 3 where it is in label 6 actually

I written this code in jupyter notebook and then converted it in .py file.

and this is the output plot.

1.0 0.9 0.8 0.7 0.6 0.5 1.4 precision recall accuracy 2 النا 4 5 7 8 9 10