Decision Trees Intro

Part 1: The Restaurant Dataset

import pandas as pd

# This will get the dataset
# It's a good practice to go ahead and download it (curl/wget)
# and change this cell to read locally

df = pd.read_csv("https://aet-cs.github.io/white/ML/lessons/restaurant.csv")

df

	Alt	Bar	Fri	Hun	Pat	Price	Rain	Res	Type	Est	Wait
0	Yes	No	No	Yes	Some	$$$	No	Yes	French	0-10	Yes
1	Yes	No	No	Yes	Full	$	No	No	Thai	30-60	No
2	No	Yes	No	No	Some	$	No	No	Burger	0-10	Yes
3	Yes	No	Yes	Yes	Full	$	No	No	Thai	10-30	Yes
4	Yes	No	Yes	No	Full	$$$	No	Yes	French	>60	No
5	No	Yes	No	Yes	Some	$$	Yes	Yes	Italian	0-10	Yes
6	No	Yes	No	No	None	$	Yes	No	Burger	0-10	No
7	No	No	No	Yes	Some	$$	Yes	Yes	Thai	0-10	Yes
8	No	Yes	Yes	No	Full	$	Yes	No	Burger	>60	No
9	Yes	Yes	Yes	Yes	Full	$$$	No	Yes	Italian	10-30	No
10	No	No	No	No	None	$	No	No	Thai	0-10	No
11	Yes	Yes	Yes	Yes	Full	$	No	No	Burger	30-60	Yes

Check out the documentation for Decision Tree Classifiers and implement one for the Restaurant dataset. Print out your decision tree and its accuracy. (It's a small dataset so using all the data for training is OK). Unfortunately the scikit-learn tree classifiers require numberical data so we will label encode our dataset first. (One hot encoding is also a possibility).

from sklearn import preprocessing
le = preprocessing.LabelEncoder()
for c in df.columns:
    le.fit(df[c])
    df[c] = le.transform(df[c])

## Create your training X and y (you can use the whole dataset)
## use scikit-learn to make a decision tree
## calculate its accuracy and metrics

Because of the required encoding, and renaming of features, it's not easy to interpret this tree and compare it to the ones we made in class. Nevertheless, it's a good little example of how to make a decision tree in scikit-learn. In a later notebook we'll look at a tree made using the algorithm from class.

Part 2: The entropy of English

Install nltk (natural language toolkit) following the commands below.

!pip install nltk

The next cell will open an interactive window (which is a bit weird). Follow the prompts to download a library called 'brown'

import nltk

## delete the next line after you download "brown" (or comment it)
nltk.download()

brown.words is a list of words

from nltk.corpus import brown

brown.words()

len(brown.words())

Your job is to use these words to compute, using standard python, the entropy of the English language. Only consider 27 characters -- the alphabet plus space.

# your code!