Due Thursday, October 17, 11:59pm
In this assignment, you'll be analyzing movie reviews from Rotten Tomatoes. This assignment will cover:
Useful libraries for this assignment
In [207]:
%matplotlib inline
import json
import requests
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
pd.set_option('display.width', 500)
pd.set_option('display.max_columns', 30)
# set some nicer defaults for matplotlib
from matplotlib import rcParams
#these colors come from colorbrewer2.org. Each is an RGB triplet
dark2_colors = [(0.10588235294117647, 0.6196078431372549, 0.4666666666666667),
(0.8509803921568627, 0.37254901960784315, 0.00784313725490196),
(0.4588235294117647, 0.4392156862745098, 0.7019607843137254),
(0.9058823529411765, 0.1607843137254902, 0.5411764705882353),
(0.4, 0.6509803921568628, 0.11764705882352941),
(0.9019607843137255, 0.6705882352941176, 0.00784313725490196),
(0.6509803921568628, 0.4627450980392157, 0.11372549019607843),
(0.4, 0.4, 0.4)]
rcParams['figure.figsize'] = (10, 6)
rcParams['figure.dpi'] = 150
rcParams['axes.color_cycle'] = dark2_colors
rcParams['lines.linewidth'] = 2
rcParams['axes.grid'] = False
rcParams['axes.facecolor'] = 'white'
rcParams['font.size'] = 14
rcParams['patch.edgecolor'] = 'none'
def remove_border(axes=None, top=False, right=False, left=True, bottom=True):
"""
Minimize chartjunk by stripping out unnecessary plot borders and axis ticks
The top/right/left/bottom keywords toggle whether the corresponding plot border is drawn
"""
ax = axes or plt.gca()
ax.spines['top'].set_visible(top)
ax.spines['right'].set_visible(right)
ax.spines['left'].set_visible(left)
ax.spines['bottom'].set_visible(bottom)
#turn off all ticks
ax.yaxis.set_ticks_position('none')
ax.xaxis.set_ticks_position('none')
#now re-enable visibles
if top:
ax.xaxis.tick_top()
if bottom:
ax.xaxis.tick_bottom()
if left:
ax.yaxis.tick_left()
if right:
ax.yaxis.tick_right()
In [208]:
pd.version.version
Out[208]:
Rotten Tomatoes gathers movie reviews from critics. An entry on the website typically consists of a short quote, a link to the full review, and a Fresh/Rotten classification which summarizes whether the critic liked/disliked the movie.
When critics give quantitative ratings (say 3/4 stars, Thumbs up, etc.), determining the Fresh/Rotten classification is easy. However, publications like the New York Times don't assign numerical ratings to movies, and thus the Fresh/Rotten classification must be inferred from the text of the review itself.
This basic task of categorizing text has many applications. All of the following questions boil down to text classification:
Language is incredibly nuanced, and there is an entire field of computer science dedicated to the topic (Natural Language Processing). Nevertheless, we can construct basic language models using fairly straightforward techniques.
You will be starting with a database of Movies, derived from the MovieLens dataset. This dataset includes information for about 10,000 movies, including the IMDB id for each movie.
Your first task is to download Rotten Tomatoes reviews from 3000 of these movies, using the Rotten Tomatoes API (Application Programming Interface).
Web APIs are a more convenient way for programs to interact with websites. Rotten Tomatoes has a nice API that gives access to its data in JSON format.
To use this, you will first need to register for an API key. For "application URL", you can use anything -- it doesn't matter.
After you have a key, the documentation page shows the various data you can fetch from Rotten Tomatoes -- each type of data lives at a different web address. The basic pattern for fetching this data with Python is as follows (compare this to the Movie Reviews tab on the documentation page):
In [209]:
api_key = 'en3yzpn423n4q9ppmysy49yq'
movie_id = '770672122' # toy story 3
url = 'http://api.rottentomatoes.com/api/public/v1.0/movies/%s/reviews.json' % movie_id
#these are "get parameters"
options = {'review_type': 'top_critic', 'page_limit': 20, 'page': 1, 'apikey': api_key}
data = requests.get(url, params=options).text
data = json.loads(data) # load a json string into a collection of lists and dicts
print json.dumps(data['reviews'][0], indent=2) # dump an object into a json string
#data
In [227]:
from io import StringIO
movie_txt = requests.get('https://raw.github.com/cs109/cs109_data/master/movies.dat').text
movie_file = StringIO(movie_txt) # treat a string like a file
movies = pd.read_csv(movie_file, delimiter='\t')
#print the first row
movies[['id', 'title', 'imdbID', 'year']].irow(0)
movies.head()
Out[227]:
In [211]:
movies[['id', 'title', 'imdbID', 'year']].irow(0)
Out[211]:
In [212]:
movies.irow(0)['id']
Out[212]:
We'd like you to write a function that looks up the first 20 Top Critic Rotten Tomatoes reviews for a movie in the movies dataframe. This involves two steps:
Movie Alias API to look up the Rotten Tomatoes movie id from the IMDB idMovie Reviews API to fetch the first 20 top-critic reviews for this movieNot all movies have Rotten Tomatoes IDs. In these cases, your function should return None. The detailed spec is below. We are giving you some freedom with how you implement this, but you'll probably want to break this task up into several small functions.
Hint In some situations, the leading 0s in front of IMDB ids are important. IMDB ids have 7 digits
In [142]:
"""
Function
--------
fetch_reviews(movies, row)
Use the Rotten Tomatoes web API to fetch reviews for a particular movie
Parameters
----------
movies : DataFrame
The movies data above
row : int
The row of the movies DataFrame to use
Returns
-------
If you can match the IMDB id to a Rotten Tomatoes ID:
A DataFrame, containing the first 20 Top Critic reviews
for the movie. If a movie has less than 20 total reviews, return them all.
This should have the following columns:
critic : Name of the critic
fresh : 'fresh' or 'rotten'
imdb : IMDB id for the movie
publication: Publication that the critic writes for
quote : string containing the movie review quote
review_data: Date of review
rtid : Rotten Tomatoes ID for the movie
title : Name of the movie
If you cannot match the IMDB id to a Rotten Tomatoes ID, return None
Examples
--------
>>> reviews = fetch_reviews(movies, 0)
>>> print len(reviews)
20
>>> print reviews.irow(1)
critic Derek Adams
fresh fresh
imdb 114709
publication Time Out
quote So ingenious in concept, design and execution ...
review_date 2009-10-04
rtid 9559
title Toy story
Name: 1, dtype: object
"""
from pandas.io.json import json_normalize
def fetch_reviews(movies, row):
movie_id = str(movies.irow(row)['imdbID']).zfill(7)
url = 'http://api.rottentomatoes.com/api/public/v1.0/movie_alias.json?id=%s&type=imdb&apikey=en3yzpn423n4q9ppmysy49yq' % movie_id
data = requests.get(url).text
data = json.loads(data)
if not data.has_key('error'):
rt_id = data['id']
url2 = 'http://api.rottentomatoes.com/api/public/v1.0/movies/%s/reviews.json?review_type=top_critic&page_limit=20&page=1&country=us&apikey=en3yzpn423n4q9ppmysy49yq' %rt_id
data2 = requests.get(url2).text
data2 = json.loads(data2)
if data2.has_key('reviews'):
if len(data2['reviews']) > 0:
df = json_normalize(data2,'reviews')
df['title'] = data['title']
df['rtid'] = data['id']
df['imdb_title'] = movies.irow(row)['title']
df['imdb'] = movies.irow(row)['imdbID']
if 'original_score' in df.columns:
df.drop(['original_score'],inplace=True,axis=1)
if 'links' in df.columns:
df.drop(['links'],inplace=True,axis=1)
df.rename(columns={'date': 'review_date', 'freshness': 'fresh'}, inplace=True)
df = df.reindex_axis(sorted(df.columns), axis=1)
return df
else:
return None
In [289]:
#fetch_reviews(movies, 108)
Use the function you wrote to retrieve reviews for the first 3,000 movies in the movies dataframe.
In [230]:
"""
Function
--------
build_table
Parameters
----------
movies : DataFrame
The movies data above
rows : int
The number of rows to extract reviews for
Returns
--------
A dataframe
The data obtained by repeatedly calling `fetch_reviews` on the first `rows`
of `movies`, discarding the `None`s,
and concatenating the results into a single DataFrame
"""
def build_table(movies, rows):
l = list(range(rows))
pd_init = fetch_reviews(movies, 0)
l.pop(0)
for num in l:
print ("Checking Index: " + str(num))
pd_init = pd.concat([pd_init, fetch_reviews(movies, num)])
return pd_init
#build_table(movies, 3)
In [145]:
# SCRATCH SPACE
#pd_init.tail()
#pd_init.to_csv("/Users/xbsd/python/rt_movies.csv", index=False)
#pd_init.tail()
movies[movies.title=="The Closer You Get"]
movies[movies.index==2999]
z = fetch_reviews(movies, 2999)
print z
In [238]:
# SCRATCH SPACE
row = 108
movie_id = str(movies.irow(row)['imdbID']).zfill(7)
url = 'http://api.rottentomatoes.com/api/public/v1.0/movie_alias.json?id=%s&type=imdb&apikey=en3yzpn423n4q9ppmysy49yq' % movie_id
data = requests.get(url).text
data = json.loads(data)
rt_id = data['id']
url2 = 'http://api.rottentomatoes.com/api/public/v1.0/movies/%s/reviews.json?review_type=top_critic&page_limit=20&page=1&country=us&apikey=en3yzpn423n4q9ppmysy49yq' %rt_id
data2 = requests.get(url2).text
data2 = json.loads(data2)
df = json_normalize(data2,'reviews')
data['id']
df['title'] = "test"
df['title'] = data['title']
df['rtid'] = data['id']
df['imdb'] = movies.irow(row)['title']
df.drop(['links','original_score'],inplace=True,axis=1)
df.rename(columns={'date': 'review_date', 'freshness': 'fresh'}, inplace=True)
#df = df.reindex_axis(sorted(df.columns), axis=1)
#df
In [290]:
#you can toggle which lines are commented, if you
#want to re-load your results to avoid repeatedly calling this function
#critics = build_table(movies, 3000)
#critics.to_csv('critics.csv', index=False)
critics = pd.read_csv('/Users/xbsd/python/rt_movies.csv')
critics.drop(['imdb_title'],inplace=True,axis=1)
#for this assignment, let's drop rows with missing data
critics = critics[-critics.quote.isnull()]
#critics = critics[critics.fresh != 'none']
#critics = critics[critics.quote.str.len() > 0]
In [291]:
#critics.dropna()
#critics[-critics.quote.isnull()].shape
A quick sanity check that everything looks ok at this point
In [239]:
assert set(critics.columns) == set('critic fresh imdb publication '
'quote review_date rtid title'.split())
assert len(critics) > 10000
2.1 How many reviews, critics, and movies are in this dataset?
In [292]:
#your code here
num_reviews = len(critics)
num_critics = len(critics.critic.unique())
num_movies = len(critics.rtid.unique())
print ("Num Reviews: " + str(num_reviews) + " Num Critics: " + str(num_critics) + " Num Movies: " + str(num_movies))
2.2 What does the distribution of number of reviews per reviewer look like? Make a histogram
In [293]:
#Your code here
def histogram_style():
remove_border(left=False)
plt.grid(False)
plt.grid(axis='y', color='w', linestyle='-', lw=1)
critics.groupby('critic').rtid.count().hist(log=True, bins=range(20), edgecolor='white')
plt.xlabel("Number of reviews per critic")
plt.ylabel("N")
histogram_style()
In [241]:
#Your code here
#critics.critic.hist(figsize=(10,10))
critics.critic.value_counts().plot(kind='bar')
Out[241]:
2.3 List the 5 critics with the most reviews, along with the publication they write for
In [242]:
#Your code here
gb = critics.groupby(["critic"])
z = gb.agg({'critic':np.count_nonzero,'publication':np.unique})
top_critics = z.sort(columns="critic", ascending=False)[0:5]
z2 = gb.agg({'critic':np.unique,'publication':np.unique})
top_critics['criticname']=top_critics.index.values
print z2[z2['critic'].isin(list(top_critics.criticname))]
top_critics
Out[242]:
2.4 Of the critics with > 100 reviews, plot the distribution of average "freshness" rating per critic
In [243]:
#Your code here
gb = critics.groupby(["critic"])
z = gb.agg({'critic':np.count_nonzero,'publication':np.unique, \
'fresh': lambda x: sum(x=="fresh")})
z['criticname'] = z.index.values
z['freshness'] = z.fresh/z.critic
top_critics = z.sort(columns="critic", ascending=False)
top_critics['criticname'] = top_critics.index.values
li = top_critics.query('critic > 100').index.values
res = z[z['criticname'].isin(list(li))]
#res.sort(columns="freshness",inplace=True)
res.plot(x="criticname", y="freshness", rot=90)
Out[243]:
2.5
Using the original movies dataframe, plot the rotten tomatoes Top Critics Rating as a function of year. Overplot the average for each year, ignoring the score=0 examples (some of these are missing data). Comment on the result -- is there a trend? What do you think it means?
In [244]:
sub = movies[['rtTopCriticsRating', 'year']]
sub = sub[(sub.rtTopCriticsRating.values != "\N")]
sub[['rtTopCriticsRating']] = sub[['rtTopCriticsRating']].astype('float')
gb2 = sub.groupby("year")
sub2 = gb2.agg({'rtTopCriticsRating':np.mean})
plt.scatter(x=sub.year, y=sub.rtTopCriticsRating.values, c='r', alpha=0.5)
plt.plot(sub2.index, sub2.rtTopCriticsRating,'r-', c='b', alpha=0.8)
Out[244]:
Your Comment Here
You will now use a Naive Bayes classifier to build a prediction model for whether a review is fresh or rotten, depending on the text of the review. See Lecture 9 for a discussion of Naive Bayes.
Most models work with numerical data, so we need to convert the textual collection of reviews to something numerical. A common strategy for text classification is to represent each review as a "bag of words" vector -- a long vector of numbers encoding how many times a particular word appears in a blurb.
Scikit-learn has an object called a CountVectorizer that turns text into a bag of words. Here's a quick tutorial:
In [245]:
from sklearn.feature_extraction.text import CountVectorizer
text = ['Hop on pop', 'Hop off pop', 'Hop Hop hop']
print "Original text is\n", '\n'.join(text)
vectorizer = CountVectorizer(min_df=0)
# call `fit` to build the vocabulary
vectorizer.fit(text)
# call `transform` to convert text to a bag of words
x = vectorizer.transform(text)
# CountVectorizer uses a sparse array to save memory, but it's easier in this assignment to
# convert back to a "normal" numpy array
x = x.toarray()
print
print "Transformed text vector is \n", x
# `get_feature_names` tracks which word is associated with each column of the transformed x
print
print "Words for each feature:"
print vectorizer.get_feature_names()
# Notice that the bag of words treatment doesn't preserve information about the *order* of words,
# just their frequency
3.1
Using the critics dataframe, compute a pair of numerical X, Y arrays where:
(nreview, nwords) array. Each row corresponds to a bag-of-words representation for a single review. This will be the input to your model.nreview-element 1/0 array, encoding whether a review is Fresh (1) or Rotten (0). This is the desired output from your model.
In [246]:
#hint: Consult the scikit-learn documentation to
# learn about what these classes do do
from sklearn.cross_validation import train_test_split
from sklearn.naive_bayes import MultinomialNB
"""
Function
--------
make_xy
Build a bag-of-words training set for the review data
Parameters
-----------
critics : Pandas DataFrame
The review data from above
vectorizer : CountVectorizer object (optional)
A CountVectorizer object to use. If None,
then create and fit a new CountVectorizer.
Otherwise, re-fit the provided CountVectorizer
using the critics data
Returns
-------
X : numpy array (dims: nreview, nwords)
Bag-of-words representation for each review.
Y : numpy array (dims: nreview)
1/0 array. 1 = fresh review, 0 = rotten review
Examples
--------
X, Y = make_xy(critics)
"""
def make_xy(critics, vectorizer=None):
#Your code here
if vectorizer == None:
vectorizer = CountVectorizer(min_df=0)
text = critics.quote.ravel()
vectorizer.fit(text)
X = vectorizer.transform(text)
X = X.toarray()
Y = np.array(1 * (critics.fresh=="fresh"))
return (X, Y)
In [247]:
X, Y = make_xy(critics, vectorizer = CountVectorizer(min_df = best_min_df))
In [248]:
np.sum(Y - (1 * (critics.fresh=="fresh")))
Out[248]:
3.2 Next, randomly split the data into two groups: a training set and a validation set.
Use the training set to train a MultinomialNB classifier,
and print the accuracy of this model on the validation set
Hint
You can use train_test_split to split up the training data
In [250]:
#Your code here
X_train, X_test, y_train, y_test = train_test_split(X, Y)
clf = MultinomialNB()
clf.fit(X_train, y_train)
predicted_train = clf.predict(X_train)
predicted_test = clf.predict(X_test)
trains=X_train.reshape(1,-1).flatten()
tests=X_test.reshape(1,-1).flatten()
In [251]:
X_test
Out[251]:
In [252]:
X.shape
Out[252]:
In [253]:
'''
>>> import numpy as np
>>> from sklearn.metrics import accuracy_score
>>> y_pred = [0, 2, 1, 3]
>>> y_true = [0, 1, 2, 3]
>>> accuracy_score(y_true, y_pred)
0.5
>>> accuracy_score(y_true, y_pred, normalize=False)
2
'''
from sklearn.metrics import accuracy_score
accuracy_train = accuracy_score(y_train, predicted_train)
accuracy_test = accuracy_score(y_test, predicted_test)
print ("Train Accuracy: " + str(accuracy_train) + " Test Accuracy: " + str(accuracy_test))
3.3:
We say a model is overfit if it performs better on the training data than on the test data. Is this model overfit? If so, how much more accurate is the model on the training data compared to the test data?
In [254]:
# Your code here. Print the accuracy on the test and training dataset
print ("Train Accuracy: " + str(accuracy_train) + " Test Accuracy: " + str(accuracy_test))
print ("Training Accuracy is better than on the Test Set by " + str(accuracy_train - accuracy_test))
In [255]:
'''
In [120]: rows = random.sample(df.index, 10)
In [121]: df_10 = df.ix[rows]
'''
# Select Random Samples
fraction = 0.75 # Change this for your examples !!!
num = np.round(len(X)*fraction).astype(int)
z = range(len(X))
ind = np.random.choice(z, num, replace=False)
samp_X = X[ind]
samp_Y = Y[ind]
samp_X_proba = clf.predict_proba(samp_X)
In [255]:
Interpret these numbers in a few sentences here
3.4: Model Calibration
Bayesian models like the Naive Bayes classifier have the nice property that they compute probabilities of a particular classification -- the predict_proba and predict_log_proba methods of MultinomialNB compute these probabilities.
Being the respectable Bayesian that you are, you should always assess whether these probabilities are calibrated -- that is, whether a prediction made with a confidence of x% is correct approximately x% of the time. We care about calibration because it tells us whether we can trust the probabilities computed by a model. If we can trust model probabilities, we can make better decisions using them (for example, we can calculate how much we should bet or invest in a given prediction).
Let's make a plot to assess model calibration. Schematically, we want something like this:
In words, we want to:
clf.predict_probaHints
The output of clf.predict_proba(X) is a (N example, 2) array. The first column gives the probability $P(Y=0)$ or $P(Rotten)$, and the second gives $P(Y=1)$ or $P(Fresh)$.
The above image is just a guideline -- feel free to explore other options!
In [256]:
df = pd.DataFrame({'samp_X_proba_1':samp_X_proba[:,1], 'samp_Y':samp_Y})
#df.sort(columns="samp_X_proba_1",inplace=True)
bin1 = np.arange(0, 101, 5)/100.
df['bin_num'] = np.digitize(df.samp_X_proba_1, bin1)
gb = df.groupby('bin_num')
gb_agg = gb.agg({'samp_Y':lambda x: np.sum(x == 1), 'bin_num':np.count_nonzero})
gb_agg['fresh_pct'] = gb_agg.samp_Y/gb_agg.bin_num
gb_agg['uncertainty'] = np.sqrt(gb_agg.fresh_pct * (1 - gb_agg.fresh_pct) / gb_agg.bin_num)
gb_agg
plt.plot(gb_agg.index.values, gb_agg.fresh_pct,'ro-')
plt.plot(np.arange(21), np.arange(21)/20.)
Out[256]:
In [257]:
plt.hist(df.bin_num, bins=20,rwidth=0.9)
Out[257]:
In [258]:
X
Out[258]:
In [259]:
"""
Function
--------
calibration_plot
Builds a plot like the one above, from a classifier and review data
Inputs
-------
clf : Classifier object
A MultinomialNB classifier
X : (Nexample, Nfeature) array
The bag-of-words data
Y : (Nexample) integer array
1 if a review is Fresh
"""
#your code here
def calibration_plot(clf, X, Y):
# Prepare the Data
fraction = 1.0 # Change this for your examples !!!
num = np.round(len(X)*fraction).astype(int)
z = range(len(X))
ind = np.random.choice(z, num, replace=False)
samp_X = X[ind]
samp_Y = Y[ind]
samp_X_proba = clf.predict_proba(samp_X)
# Prepare the DataFrame
df = pd.DataFrame({'samp_X_proba_1':samp_X_proba[:,1], 'samp_Y':samp_Y})
#df.sort(columns="samp_X_proba_1",inplace=True)
bin1 = np.arange(0, 101, 5)/100.
df['bin_num'] = np.digitize(df.samp_X_proba_1, bin1)
gb = df.groupby('bin_num')
gb_agg = gb.agg({'samp_Y':lambda x: np.sum(x == 1), 'bin_num':np.count_nonzero})
gb_agg['fresh_pct'] = gb_agg.samp_Y/gb_agg.bin_num
gb_agg['uncertainty'] = np.sqrt(gb_agg.fresh_pct * (1 - gb_agg.fresh_pct) / gb_agg.bin_num)
# gb_agg
# Plot Figures
plt.figure(0)
plt.plot(gb_agg.index.values, gb_agg.fresh_pct,'ro-')
plt.plot(np.arange(21), np.arange(21)/20.)
plt.figure(1)
plt.hist(df.bin_num, bins=20,rwidth=0.9)
plt.show
return None
In [260]:
calibration_plot(clf, X_test, y_test)
3.5 We might say a model is over-confident if the freshness fraction is usually closer to 0.5 than expected (that is, there is more uncertainty than the model predicted). Likewise, a model is under-confident if the probabilities are usually further away from 0.5. Is this model generally over- or under-confident?
Your Answer Here
Our classifier has a few free parameters. The two most important are:
The min_df keyword in CountVectorizer, which will ignore words which appear in fewer than min_df fraction of reviews. Words that appear only once or twice can lead to overfitting, since words which occur only a few times might correlate very well with Fresh/Rotten reviews by chance in the training dataset.
The alpha keyword in the Bayesian classifier is a "smoothing parameter" -- increasing the value decreases the sensitivity to any single feature, and tends to pull prediction probabilities closer to 50%.
As discussed in lecture and HW2, a common technique for choosing appropriate values for these parameters is cross-validation. Let's choose good parameters by maximizing the cross-validated log-likelihood.
3.6 Using clf.predict_log_proba, write a function that computes the log-likelihood of a dataset
In [261]:
"""
Function
--------
log_likelihood
Compute the log likelihood of a dataset according to a bayesian classifier.
The Log Likelihood is defined by
L = Sum_fresh(logP(fresh)) + Sum_rotten(logP(rotten))
Where Sum_fresh indicates a sum over all fresh reviews,
and Sum_rotten indicates a sum over rotten reviews
Parameters
----------
clf : Bayesian classifier
x : (nexample, nfeature) array
The input data
y : (nexample) integer array
Whether each review is Fresh
"""
#your code here
def log_likelihood(clf, x, y):
x_proba = clf.predict_log_proba(x)
prob_rotten = x_proba[:,0]
prob_fresh = x_proba[:,1]
df = pd.DataFrame({'y':y, 'logP_fresh':prob_fresh, 'logP_rotten':prob_rotten})
df['y']=df.y.astype("str")
gb = df.groupby('y')
res = gb.agg({'y':np.count_nonzero, 'logP_fresh': np.sum, 'logP_rotten': np.sum})
logP_fresh = res[res.index=="1"][['logP_fresh']].values
logP_rotten = res[res.index=="0"][['logP_rotten']].values
L = (logP_fresh + logP_rotten).ravel()
#print ("LogP Fresh: " + str(logP_fresh) + " LogP Rotten: " + \
# str(logP_rotten) + " L: " + str(L))
return L
#x = log_likelihood(clf, X_test, y_test)
#x
In [262]:
x < 10
Out[262]:
Here's a function to estimate the cross-validated value of a scoring function, given a classifier and data
In [263]:
from sklearn.cross_validation import KFold
def cv_score(clf, x, y, score_func):
"""
Uses 5-fold cross validation to estimate a score of a classifier
Inputs
------
clf : Classifier object
x : Input feature vector
y : Input class labels
score_func : Function like log_likelihood, that takes (clf, x, y) as input,
and returns a score
Returns
-------
The average score obtained by randomly splitting (x, y) into training and
test sets, fitting on the training set, and evaluating score_func on the test set
Examples
cv_score(clf, x, y, log_likelihood)
"""
result = 0
nfold = 5
for train, test in KFold(y.size, nfold): # split data into train/test groups, 5 times
clf.fit(x[train], y[train]) # fit
result += score_func(clf, x[test], y[test]) # evaluate score function on held-out data
return result / nfold # average
# --- Very Important -- Existing Functionality ---
# as a side note, this function is builtin to the newest version of sklearn. We could just write
# sklearn.cross_validation.cross_val_score(clf, x, y, scorer=log_likelihood).
# ------------------------------------------------
# cv_score(clf, X, Y, log_likelihood)
3.7
Fill in the remaining code in this block, to loop over many values of alpha and min_df to determine
which settings are "best" in the sense of maximizing the cross-validated log-likelihood
In [264]:
#the grid of parameters to search over
alphas = [0, .1, 1, 5, 10, 50]
min_dfs = [1e-5, 1e-4, 1e-3, 1e-2, 1e-1]
# alphas = [.1] # For Testing
# min_dfs = [1e-5] # For Testing
#Find the best value for alpha and min_df, and the best classifier
best_alpha = None
best_min_df = None
max_loglike = -np.inf
i = 0
iterations = len(alphas) * len(min_dfs)
for alpha in alphas:
for min_df in min_dfs:
print ("Starting Iteration: " + str(i) + " of " + str(iterations))
print ("Alpha: " + str(alpha) + " Min DF: " + str(min_df))
vectorizer = CountVectorizer(min_df = min_df)
X, Y = make_xy(critics, vectorizer)
#your code here
clf = MultinomialNB(alpha=alpha)
new_score = cv_score(clf, X, Y, log_likelihood)
if new_score > max_loglike:
max_loglike = new_score
best_alpha = alpha
best_min_df = min_df
print ("LogLik Score = " + str(new_score))
print (" ---")
i = i + 1
In [265]:
print "alpha: %f" % best_alpha
print "min_df: %f" % best_min_df
3.8 Now that you've determined values for alpha and min_df that optimize the cross-validated log-likelihood, repeat the steps in 3.1, 3.2, and 3.4 to train a final classifier with these parameters, re-evaluate the accuracy, and draw a new calibration plot.
In [266]:
#Your code here
X, Y = make_xy(critics, vectorizer = CountVectorizer(min_df = best_min_df))
X_train, X_test, y_train, y_test = train_test_split(X, Y)
clf = MultinomialNB(alpha = best_alpha)
clf.fit(X_train, y_train)
predicted_train = clf.predict(X_train)
predicted_test = clf.predict(X_test)
trains=X_train.reshape(1,-1).flatten()
tests=X_test.reshape(1,-1).flatten()
from sklearn.metrics import accuracy_score
accuracy_train = accuracy_score(y_train, predicted_train)
accuracy_test = accuracy_score(y_test, predicted_test)
print ("Train Accuracy: " + str(accuracy_train) + " Test Accuracy: " + str(accuracy_test))
calibration_plot(clf, X_test, y_test)
log_likelihood(clf, X_test, y_test)
cv_score(clf, X, Y, log_likelihood)
Out[266]:
In [267]:
log_likelihood(clf, X_test, y_test)
Out[267]:
In [268]:
cv_score(clf, X, Y, log_likelihood)
Out[268]:
3.9 Discuss the various ways in which Cross-Validation has affected the model. Is the new model more or less accurate? Is overfitting better or worse? Is the model more or less calibrated?
Your Answer Here
To think about/play with, but not to hand in: What would happen if you tried this again using a function besides the log-likelihood -- for example, the classification accuracy?
4.1
Using your classifier and the vectorizer.get_feature_names method, determine which words best predict a positive or negative review. Print the 10 words
that best predict a "fresh" review, and the 10 words that best predict a "rotten" review. For each word, what is the model's probability of freshness if the word appears one time?
Try computing the classification probability for a feature vector which consists of all 0s, except for a single 1. What does this probability refer to?
np.eye generates a matrix where the ith row is all 0s, except for the ith column which is 1.
In [288]:
# Your code here
def show_most_informative_features(vectorizer, clf, n=20):
feature_names = vectorizer.get_feature_names()
coefs_with_fns = sorted(zip(clf.coef_[0], feature_names))
top = zip(coefs_with_fns[:n], coefs_with_fns[:-(n + 1):-1])
for (coef_1, fn_1), (coef_2, fn_2) in top:
print "\t%.4f \t%-15s \t\t %.4f \t% -15s" % (coef_1, fn_1, coef_2, fn_2)
n = 10
feature_names = vectorizer.get_feature_names()
coefs_with_fns = sorted(zip(clf.coef_, feature_names))
top = zip(coefs_with_fns[:n], coefs_with_fns[:-(n + 1):-1])
show_most_informative_features(vectorizer, clf, n=20)
#coefs_with_fns[:]
top
words = np.array(vectorizer.get_feature_names())
x = np.eye(X_test.shape[1])
probs = clf.predict_log_proba(x)[:, 0]
ind = np.argsort(probs)
[ind[:10]]
vectorizer.get_feature_names()
Out[288]:
In [270]:
def print_top10(vectorizer, clf, class_labels):
"""Prints features with the highest coefficient values, per class"""
feature_names = vectorizer.get_feature_names()
for i, class_label in enumerate(class_labels):
top10 = np.argsort(clf.coef_[i])[-10:]
print("%s: %s" % (class_label,
" ".join(feature_names[j] for j in top10)))
(clf.coef_.shape)
Out[270]:
4.2
One of the best sources for inspiration when trying to improve a model is to look at examples where the model performs poorly.
Find 5 fresh and rotten reviews where your model performs particularly poorly. Print each review.
In [271]:
make_xy(critics)
Out[271]:
In [272]:
critics[critics.imdb==110955]
#Y[5039:5052]
Out[272]:
In [275]:
#Your code here
#Your code here
X, Y = make_xy(critics, vectorizer = CountVectorizer(min_df = best_min_df))
X_train, X_test, y_train, y_test = train_test_split(X, Y)
clf = MultinomialNB(alpha = best_alpha)
clf.fit(X_train, y_train)
all_pred = clf.predict(X)
all_act = Y
alldata = critics
yzero = clf.predict_proba(X)[:,0]
yone = clf.predict_proba(X)[:,1]
temp = pd.DataFrame({'actual':all_act, 'pred': all_pred, 'proba0': yzero, 'proba1':yone})
temp
Out[275]:
In [279]:
comb = pd.concat([temp, critics], axis=1)
fresh = comb[comb.actual == 1]
#fresh.sort(columns="proba1", inplace=True)
print "Actual Was Fresh, But Predicted Rotten"
fresh[fresh.quote.notnull()].head()
comb
Out[279]:
In [199]:
Y[1000:1010]
critics[1000:1010]
df2 = pd.DataFrame({'Y':Y})
Out[199]:
In [ ]:
In [135]:
rotten = comb[comb.actual == 0]
rotten.sort(columns="proba0", inplace=True)
print "Actual Was Rotten, But Predicted Fresh"
rotten[rotten.quote.notnull()].head()
Out[135]:
4.3 What do you notice about these mis-predictions? Naive Bayes classifiers assume that every word affects the probability independently of other words. In what way is this a bad assumption? In your answer, report your classifier's Freshness probability for the review "This movie is not remarkable, touching, or superb in any way".
Your answer here
4.4 If this was your final project, what are 3 things you would try in order to build a more effective review classifier? What other exploratory or explanatory visualizations do you think might be helpful?
Your answer here
Restart and run your notebook one last time, to make sure the output from each cell is up to date. To submit your homework, create a folder named lastname_firstinitial_hw3 and place your solutions in the folder. Double check that the file is still called HW3.ipynb, and that it contains your code. Please do not include the critics.csv data file, if you created one. Compress the folder (please use .zip compression) and submit to the CS109 dropbox in the appropriate folder. If we cannot access your work because these directions are not followed correctly, we will not grade your work!
css tweaks in this cell