基于文本来推荐相似酒店

基于文本来推荐相似酒店

查看数据集基本信息

import pandas as pd
import numpy as np
from nltk.corpus import stopwords
from sklearn.metrics.pairwise import linear_kernel
from sklearn.feature_extraction.text import CountVectorizer 
from sklearn.feature_extraction.text import TfidfVectorizer
import re
import random
import cufflinks
import cufflinks
from plotly.offline import iplot

df=pd.read_csv("Seattle_Hotels.csv",encoding="latin-1")
df.head()

df.shape

(152, 3)

df['desc'][0]

"Located on the southern tip of Lake Union, the Hilton Garden Inn Seattle Downtown hotel is perfectly located for business and leisure. \nThe neighborhood is home to numerous major international companies including Amazon, Google and the Bill & Melinda Gates Foundation. A wealth of eclectic restaurants and bars make this area of Seattle one of the most sought out by locals and visitors. Our proximity to Lake Union allows visitors to take in some of the Pacific Northwest's majestic scenery and enjoy outdoor activities like kayaking and sailing. over 2,000 sq. ft. of versatile space and a complimentary business center. State-of-the-art A/V technology and our helpful staff will guarantee your conference, cocktail reception or wedding is a success. Refresh in the sparkling saltwater pool, or energize with the latest equipment in the 24-hour fitness center. Tastefully decorated and flooded with natural light, our guest rooms and suites offer everything you need to relax and stay productive. Unwind in the bar, and enjoy American cuisine for breakfast, lunch and dinner in our restaurant. The 24-hour Pavilion Pantry? stocks a variety of snacks, drinks and sundries."

查看酒店描述中主要介绍信息

vec=CountVectorizer().fit(df['desc'])

vec=CountVectorizer().fit(df['desc'])
bag_of_words=vec.transform(df['desc'])
sum_words=bag_of_words.sum(axis=0)
words_freq=[(word,sum_words[0,idx]) for word,idx in vec.vocabulary_.items()]
sum_words=sorted(words_freq,key= lambda x: x[1],reverse=True)
sum_words[1:10]

[('and', 1062),
 ('of', 536),
 ('seattle', 533),
 ('to', 471),
 ('in', 449),
 ('our', 359),
 ('you', 304),
 ('hotel', 295),
 ('with', 280)]

bag_of_words=vec.transform(df['desc'])
bag_of_words.shape

(152, 3200)

bag_of_words.toarray()

array([[0, 1, 0, ..., 0, 0, 0],
       [0, 0, 0, ..., 0, 0, 0],
       [0, 0, 0, ..., 0, 0, 0],
       ...,
       [0, 0, 0, ..., 0, 0, 0],
       [0, 0, 0, ..., 0, 0, 0],
       [0, 0, 0, ..., 1, 0, 0]], dtype=int64)

sum_words=bag_of_words.sum(axis=0)
sum_words

matrix([[ 1, 11, 11, ...,  2,  6,  2]], dtype=int64)

words_freq=[(word,sum_words[0,idx]) for word,idx in vec.vocabulary_.items()]

sum_words=sorted(words_freq,key= lambda x: x[1],reverse=True)

sum_words[1:10]

[('and', 1062),
 ('of', 536),
 ('seattle', 533),
 ('to', 471),
 ('in', 449),
 ('our', 359),
 ('you', 304),
 ('hotel', 295),
 ('with', 280)]

将以上信息整合成函数

def get_top_n_words(corpus,n=None):
    vec=CountVectorizer().fit(df['desc'])
    bag_of_words=vec.transform(df['desc'])
    sum_words=bag_of_words.sum(axis=0)
    words_freq=[(word,sum_words[0,idx]) for word,idx in vec.vocabulary_.items()]
    sum_words=sorted(words_freq,key= lambda x: x[1],reverse=True)
    return sum_words[:n]

common_words=get_top_n_words(df['desc'],20)
common_words

[('the', 1258),
 ('and', 1062),
 ('of', 536),
 ('seattle', 533),
 ('to', 471),
 ('in', 449),
 ('our', 359),
 ('you', 304),
 ('hotel', 295),
 ('with', 280),
 ('is', 271),
 ('at', 231),
 ('from', 224),
 ('for', 216),
 ('your', 186),
 ('or', 161),
 ('center', 151),
 ('are', 136),
 ('downtown', 133),
 ('on', 129)]

df1=pd.DataFrame(common_words,columns=['desc','count'])

common_words=get_top_n_words(df['desc'],20)
df2=pd.DataFrame(common_words,columns=['desc','count'])
chart_info2=df2.groupby(['desc']).sum().sort_values('count',ascending=False)
chart_info2.plot(kind='barh',figsize=(14,10),title='top 20 before remove stopwords')

<AxesSubplot:title={'center':'top 20 before remove stopwords'}, ylabel='desc'>    

chart_info1=df1.groupby(['desc']).sum().sort_values('count',ascending=False)

chart_info1.plot(kind='barh',figsize=(14,10),title='top 20 before remove stopwords')

<AxesSubplot:title={'center':'top 20 before remove stopwords'}, ylabel='desc'>

def get_any1_top_n_words_after_stopwords(corpus,n=None):
    vec=CountVectorizer(stop_words='english',ngram_range=(1,1)).fit(df['desc'])
    bag_of_words=vec.transform(df['desc'])
    sum_words=bag_of_words.sum(axis=0)
    words_freq=[(word,sum_words[0,idx]) for word,idx in vec.vocabulary_.items()]
    sum_words=sorted(words_freq,key= lambda x: x[1],reverse=True)
    return sum_words[:n]

common_words=get_any1_top_n_words_after_stopwords(df['desc'],20)
df2=pd.DataFrame(common_words,columns=['desc','count'])
chart_info2=df2.groupby(['desc']).sum().sort_values('count',ascending=False)
chart_info2.plot(kind='barh',figsize=(14,10),title='top 20 before after stopwords')

<AxesSubplot:title={'center':'top 20 before after stopwords'}, ylabel='desc'>

def get_any2_top_n_words_after_stopwords(corpus,n=None):
    vec=CountVectorizer(stop_words='english',ngram_range=(2,2)).fit(df['desc'])
    bag_of_words=vec.transform(df['desc'])
    sum_words=bag_of_words.sum(axis=0)
    words_freq=[(word,sum_words[0,idx]) for word,idx in vec.vocabulary_.items()]
    sum_words=sorted(words_freq,key= lambda x: x[1],reverse=True)
    return sum_words[:n]

common_words=get_any2_top_n_words_after_stopwords(df['desc'],20)
df2=pd.DataFrame(common_words,columns=['desc','count'])
chart_info2=df2.groupby(['desc']).sum().sort_values('count',ascending=False)
chart_info2.plot(kind='barh',figsize=(14,10),title='top 20 before after stopwords')

<AxesSubplot:title={'center':'top 20 before after stopwords'}, ylabel='desc'>

def get_any3_top_n_words_after_stopwords(corpus,n=None):
    vec=CountVectorizer(stop_words='english',ngram_range=(3,3)).fit(df['desc'])
    bag_of_words=vec.transform(df['desc'])
    sum_words=bag_of_words.sum(axis=0)
    words_freq=[(word,sum_words[0,idx]) for word,idx in vec.vocabulary_.items()]
    sum_words=sorted(words_freq,key= lambda x: x[1],reverse=True)
    return sum_words[:n]

common_words=get_any3_top_n_words_after_stopwords(df['desc'],20)
df2=pd.DataFrame(common_words,columns=['desc','count'])
chart_info2=df2.groupby(['desc']).sum().sort_values('count',ascending=False)
chart_info2.plot(kind='barh',figsize=(14,10),title='top 20 before after stopwords')

<AxesSubplot:title={'center':'top 20 before after stopwords'}, ylabel='desc'>

描述的一些统计信息

df=pd.read_csv("Seattle_Hotels.csv",encoding="latin-1")
df['desc'][0]

"Located on the southern tip of Lake Union, the Hilton Garden Inn Seattle Downtown hotel is perfectly located for business and leisure. \nThe neighborhood is home to numerous major international companies including Amazon, Google and the Bill & Melinda Gates Foundation. A wealth of eclectic restaurants and bars make this area of Seattle one of the most sought out by locals and visitors. Our proximity to Lake Union allows visitors to take in some of the Pacific Northwest's majestic scenery and enjoy outdoor activities like kayaking and sailing. over 2,000 sq. ft. of versatile space and a complimentary business center. State-of-the-art A/V technology and our helpful staff will guarantee your conference, cocktail reception or wedding is a success. Refresh in the sparkling saltwater pool, or energize with the latest equipment in the 24-hour fitness center. Tastefully decorated and flooded with natural light, our guest rooms and suites offer everything you need to relax and stay productive. Unwind in the bar, and enjoy American cuisine for breakfast, lunch and dinner in our restaurant. The 24-hour Pavilion Pantry? stocks a variety of snacks, drinks and sundries."

df['word_count']=df['desc'].apply(   lambda x:len(str(x).split(' '))    )
df.head()                                    

df['word_count'].plot(kind='hist',bins=50)

<AxesSubplot:ylabel='Frequency'>

文本处理

sub_replace=re.compile('[^0-9a-z#-]')

from nltk.corpus import stopwords

stopwords=set(stopwords.words('english'))

def clean_txt(text):
    text.lower()
    text=sub_replace.sub(' ',text)
    ''.join(    word   for word in text.split(' ')  if word not in stopwords               )
    return  text

df['desc_clean']=df['desc'].apply(clean_txt)

df['desc_clean'][0]

' ocated on the southern tip of  ake  nion  the  ilton  arden  nn  eattle  owntown hotel is perfectly located for business and leisure    he neighborhood is home to numerous major international companies including  mazon   oogle and the  ill    elinda  ates  oundation    wealth of eclectic restaurants and bars make this area of  eattle one of the most sought out by locals and visitors   ur proximity to  ake  nion allows visitors to take in some of the  acific  orthwest s majestic scenery and enjoy outdoor activities like kayaking and sailing  over 2 000 sq  ft  of versatile space and a complimentary business center   tate-of-the-art     technology and our helpful staff will guarantee your conference  cocktail reception or wedding is a success   efresh in the sparkling saltwater pool  or energize with the latest equipment in the 24-hour fitness center   astefully decorated and flooded with natural light  our guest rooms and suites offer everything you need to relax and stay productive   nwind in the bar  and enjoy  merican cuisine for breakfast  lunch and dinner in our restaurant   he 24-hour  avilion  antry  stocks a variety of snacks  drinks and sundries '

相似度计算

df.index

RangeIndex(start=0, stop=152, step=1)

df.head()

df.set_index('name' ,inplace=True)

df.index[:5]

Index(['Hilton Garden Seattle Downtown', 'Sheraton Grand Seattle',
       'Crowne Plaza Seattle Downtown', 'Kimpton Hotel Monaco Seattle ',
       'The Westin Seattle'],
      dtype='object', name='name')

tf=TfidfVectorizer(analyzer='word',ngram_range=(1,3),stop_words='english')#将原始文档集合转换为TF-IDF特性的矩阵。
tf

TfidfVectorizer(ngram_range=(1, 3), stop_words='english')

tfidf_martix=tf.fit_transform(df['desc_clean'])
tfidf_martix.shape

(152, 27694)

cosine_similarity=linear_kernel(tfidf_martix,tfidf_martix)
cosine_similarity.shape

(152, 152)

cosine_similarity[0]

array([1.        , 0.01354605, 0.02855898, 0.00666729, 0.02915865,
       0.01258837, 0.0190937 , 0.0152567 , 0.00689703, 0.01852763,
       0.01241924, 0.00919602, 0.01189826, 0.01234794, 0.01200711,
       0.01596218, 0.00979221, 0.04374643, 0.01138524, 0.02334485,
       0.02358692, 0.00829121, 0.00620275, 0.01700472, 0.0191396 ,
       0.02340334, 0.03193292, 0.00678849, 0.02272962, 0.0176494 ,
       0.0125159 , 0.03702338, 0.01569165, 0.02001584, 0.03656467,
       0.03189017, 0.00644231, 0.01008181, 0.02428547, 0.03327365,
       0.01367507, 0.00827835, 0.01722986, 0.04135263, 0.03315194,
       0.01529834, 0.03568623, 0.01294482, 0.03480617, 0.01447235,
       0.02563783, 0.01650068, 0.03328324, 0.01562323, 0.02703264,
       0.01315504, 0.02248426, 0.02690816, 0.00565479, 0.02899467,
       0.02900863, 0.00971019, 0.0439659 , 0.03020971, 0.02166199,
       0.01487286, 0.03182626, 0.00729518, 0.01764764, 0.01193849,
       0.02405471, 0.01408249, 0.02632335, 0.02027866, 0.01978292,
       0.04879328, 0.00244737, 0.01937539, 0.01388813, 0.02996677,
       0.00756079, 0.01429659, 0.0050572 , 0.00630326, 0.01496956,
       0.04104425, 0.00911942, 0.00259554, 0.00645944, 0.01460694,
       0.00794788, 0.00592598, 0.0090397 , 0.00532289, 0.01445326,
       0.01156657, 0.0098189 , 0.02077998, 0.0116756 , 0.02593775,
       0.01000463, 0.00533785, 0.0026153 , 0.02261775, 0.00680343,
       0.01859473, 0.03802118, 0.02078981, 0.01196228, 0.03744293,
       0.05164375, 0.00760035, 0.02627101, 0.01579335, 0.01852171,
       0.06768183, 0.01619049, 0.03544484, 0.0126264 , 0.01613638,
       0.00662941, 0.01184946, 0.01843151, 0.0012407 , 0.00687414,
       0.00873796, 0.04397665, 0.06798914, 0.00794379, 0.01098165,
       0.01520306, 0.01257289, 0.02087956, 0.01718063, 0.0292332 ,
       0.00489742, 0.03096065, 0.01163736, 0.01382631, 0.01386944,
       0.01888652, 0.02391748, 0.02814364, 0.01467017, 0.00332169,
       0.0023627 , 0.02348599, 0.00762246, 0.00390889, 0.01277579,
       0.00247891, 0.00854051])

求酒店的推荐

indices=pd.Series(df.index)
indices[:5]

0    Hilton Garden Seattle Downtown
1            Sheraton Grand Seattle
2     Crowne Plaza Seattle Downtown
3     Kimpton Hotel Monaco Seattle 
4                The Westin Seattle
Name: name, dtype: object

def recommendation(name,cosine_similarity):
    recommend_hotels=[]
    idx=indices[indices==name].index[0]
    score_series=pd.Series(cosine_similarity[idx]).sort_values(ascending=False)
    top_10_indexes=list(score_series[1:11].index)
    for i in top_10_indexes:
        recommend_hotels.append(list(df.index)[i])
    return recommend_hotels                                

recommendation('Hilton Garden Seattle Downtown',cosine_similarity)

['Staybridge Suites Seattle Downtown - Lake Union',
 'Silver Cloud Inn - Seattle Lake Union',
 'Residence Inn by Marriott Seattle Downtown/Lake Union',
 'MarQueen Hotel',
 'The Charter Hotel Seattle, Curio Collection by Hilton',
 'Embassy Suites by Hilton Seattle Tacoma International Airport',
 'SpringHill Suites Seattle\xa0Downtown',
 'Courtyard by Marriott Seattle Downtown/Pioneer Square',
 'The Loyal Inn',
 'EVEN Hotel Seattle - South Lake Union']