def run(path):
# Assuming that all the xml files are present in your current working directory
folderList = os.listdir(path)
# Read in all the xml files - other files such as .py files will be not read in. Extract all aspects to be
# visualized as wordclouds
asp_sent_list = []
for file_name in folderList:
ext = os.path.splitext(file_name)[-1].lower()
if ext == ".xml":
aspects = asp.get_aspect(path + "/" + file_name)
asp_cat = aspc.get_aspect_categories(aspects)
# asp_sent = asps.get_aspect_sentiment(aspects, path + "/" + file_name, categories = False)
if "food" in asp_cat:
temp_aspects = aspc.get_aspect_by_category(aspects, "food")
asp_sent = asps.get_aspect_sentiment(temp_aspects, path + "/" + file_name, categories = False)
asp_sent_list.append([file_name[:-4], temp_aspects, asp_sent])
# Read in the csv file in question. This is done to match the review ids to the restaurant id
# because we want to summarize aspects by restaurant
test = pd.read_csv("charlotte_top10.csv")
# Create a list with restaurant_ids aspects along with sentimentss
rev_rest_list = []
for i, r in test.iterrows():
for asl in asp_sent_list:
if r['review_id']==asl[0]:
rev_rest_list.append([r['restaurant_id'], asl[2] ])
# Convert list to dataframe for easy handling
rev_rest = pd.DataFrame(rev_rest_list, columns = ['rest_id','Aspect_Sent'])
# Create groups by restaurant id
grouped = rev_rest.groupby('rest_id')
# Process data so that frequency of each aspect-sentiment pair is counted
dict_check = []
temp1 = []
for g in grouped:
h = g[1]
temp = []
for i,r in h.iterrows():
c = Counter( item for item in r.Aspect_Sent.items())
d = dict(c)
temp.append(d)
dict_check = functools.reduce(operator.add, map(collections.Counter, temp))
temp1.append([g[0], dict(dict_check)])
# Create an empty dataframe with the structure which tableau requires
columns = ['rest_id', 'Aspect', 'Freq', 'Type']
temp_df = pd.DataFrame(columns = columns)
# Fill in the empty dataframe with the list created, separating out negative and positive sentiments by
# creating a column called "Type" and also storing their frequency in the column "Freq"
for t in temp1:
for k,v in t[1].items():
if k[1] == 'Positive':
temp_df = temp_df.append({'rest_id': t[0], 'Aspect':k[0], 'Freq': v, 'Type':"Positive"}, ignore_index = True)
elif k[1] == 'Negative':
temp_df = temp_df.append({'rest_id': t[0], 'Aspect':k[0], 'Freq': v, 'Type': "Negative"}, ignore_index = True)
temp_df.to_csv("result/question3.csv", encoding = "UTF-8")
def run(path, file_name, time_grain):
folderList = os.listdir(path)
aspect_list = []
# Read in the csv file in question
test = pd.read_csv(file_name)
# Create two fields of "year" and "month" which will store the extracted year and month
# from the date
test_check = test
test_check['year'] = pd.DatetimeIndex(test_check['date']).year
test_check['month'] = pd.DatetimeIndex(test_check['date']).month
# Assuming all your xml files are present in the same path, extract all aspects from the reviews and store
# them in a list
for file_name in folderList:
ext = os.path.splitext(file_name)[-1].lower()
if ext == ".xml":
print("Processing ..." + str(file_name[:-4]))
aspects = asp.get_aspect(path + "/" + file_name)
asp_cat = aspc.get_aspect_categories(aspects)
asp_sent = asps.get_aspect_sentiment(aspects,
path + "/" + file_name,
categories=True)
aspect_list.append([file_name[:-4], asp_cat, asp_sent])
# Create an empty dataframe with the below structure
columns = [
'rest_id', 'rev_id', 'year', 'month', 'overall_sentiment', 'food',
'price', 'service', 'ambience', 'others'
]
vis_df = pd.DataFrame(columns=columns)
# We have to start attaching the reviews to restaurants, create an empty list to do that
tot_list = []
# Populate the list
for i, r in test_check.iterrows():
for al in aspect_list:
if r['review_id'] == al[0]:
tot_list.append([
r['restaurant_id'], r['review_id'], r['year'], r['month'],
r['Sentiment'], al[2]
])
# Populate the list
for l in tot_list:
food_val = check(l, 'food')
price_val = check(l, 'price')
service_val = check(l, 'service')
ambience_val = check(l, 'ambience')
others_val = check(l, 'others')
vis_df = vis_df.append(
{
'rest_id': l[0],
'rev_id': l[1],
'year': l[2],
'month': l[3],
'overall_sentiment': l[4],
'food': food_val,
'price': price_val,
'service': service_val,
'ambience': ambience_val,
'others': others_val
},
ignore_index=True)
# Create an empty dataframe with the below structure and which will hold actual sentiment positive to
# overall ratios
columns = [
'rest_id', time_grain, 'overall', 'food', 'price', 'service',
'ambience', 'others'
]
rest_df = pd.DataFrame(columns=columns)
# Group your initial dataframe by restaurant id
grouped = vis_df.groupby('rest_id')
# Looping through each restaurant
for g in grouped:
h = g[1]
# Grouping by time_grain (example month or year)
g_child = h.groupby(time_grain)
# Looping through each time_grain (example month or year)
for gc in g_child:
h1 = gc[1]
# Calculating all ratios
g_over = h1.groupby('overall_sentiment')
over_ratio = get_group_size(g_over, "positive") / float(
h1.shape[0])
g_food = h1.groupby('food')
food_ratio = get_group_size(g_food, "Positive") / float(
h1.shape[0])
g_price = h1.groupby('price')
price_ratio = get_group_size(g_price, "Positive") / float(
h1.shape[0])
g_service = h1.groupby("service")
service_ratio = get_group_size(g_service, "Positive") / float(
h1.shape[0])
g_ambience = h1.groupby("ambience")
ambience_ratio = get_group_size(g_ambience, "Positive") / float(
h1.shape[0])
g_others = h1.groupby("others")
others_ratio = get_group_size(g_others, "Positive") / float(
h1.shape[0])
rest_df = rest_df.append(
{
'rest_id': g[0],
time_grain: gc[0],
'overall': over_ratio,
'food': food_ratio,
'price': price_ratio,
'service': service_ratio,
'ambience': ambience_ratio,
'others': others_ratio
},
ignore_index=True)
# Writing to file
rest_df.to_csv("result/question2.csv")
def run(path, file_name, time_grain):
folderList = os.listdir(path)
aspect_list = []
# Read in the csv file in question
test = pd.read_csv(file_name)
# Create two fields of "year" and "month" which will store the extracted year and month
# from the date
test_check = test
test_check['year'] = pd.DatetimeIndex(test_check['date']).year
test_check['month'] = pd.DatetimeIndex(test_check['date']).month
# Assuming all your xml files are present in the same path, extract all aspects from the reviews and store
# them in a list
for file_name in folderList:
ext = os.path.splitext(file_name)[-1].lower()
if ext == ".xml":
print("Processing ..." + str(file_name[:-4]))
aspects = asp.get_aspect(path + "/" + file_name)
asp_cat = aspc.get_aspect_categories(aspects)
asp_sent = asps.get_aspect_sentiment(aspects, path + "/" + file_name, categories = True)
aspect_list.append([file_name[:-4], asp_cat, asp_sent])
# Create an empty dataframe with the below structure
columns = ['rest_id', 'rev_id', 'year','month','overall_sentiment','food','price','service','ambience','others']
vis_df = pd.DataFrame(columns=columns)
# We have to start attaching the reviews to restaurants, create an empty list to do that
tot_list = []
# Populate the list
for i, r in test_check.iterrows():
for al in aspect_list:
if r['review_id'] == al[0]:
tot_list.append([r['restaurant_id'],r['review_id'],r['year'],r['month'], r['Sentiment'], al[2] ])
# Populate the list
for l in tot_list:
food_val = check(l,'food')
price_val = check(l,'price')
service_val = check(l,'service')
ambience_val = check(l,'ambience')
others_val = check(l,'others')
vis_df = vis_df.append({'rest_id': l[0], 'rev_id':l[1], 'year':l[2], 'month':l[3],'overall_sentiment':l[4], 'food':food_val, 'price':price_val, 'service':service_val, 'ambience':ambience_val, 'others':others_val }, ignore_index=True)
# Create an empty dataframe with the below structure and which will hold actual sentiment positive to
# overall ratios
columns = ['rest_id', time_grain,'overall', 'food','price','service','ambience','others']
rest_df = pd.DataFrame(columns=columns)
# Group your initial dataframe by restaurant id
grouped = vis_df.groupby('rest_id')
# Looping through each restaurant
for g in grouped:
h = g[1]
# Grouping by time_grain (example month or year)
g_child = h.groupby(time_grain)
# Looping through each time_grain (example month or year)
for gc in g_child:
h1 = gc[1]
# Calculating all ratios
g_over = h1.groupby('overall_sentiment')
over_ratio = get_group_size(g_over, "positive")/float(h1.shape[0])
g_food = h1.groupby('food')
food_ratio = get_group_size(g_food, "Positive")/float(h1.shape[0])
g_price = h1.groupby('price')
price_ratio = get_group_size(g_price, "Positive")/float(h1.shape[0])
g_service = h1.groupby("service")
service_ratio = get_group_size(g_service, "Positive")/float(h1.shape[0])
g_ambience = h1.groupby("ambience")
ambience_ratio = get_group_size(g_ambience, "Positive")/float(h1.shape[0])
g_others = h1.groupby("others")
others_ratio = get_group_size(g_others, "Positive")/float(h1.shape[0])
rest_df = rest_df.append({'rest_id': g[0],time_grain:gc[0], 'overall':over_ratio, 'food':food_ratio, 'price':price_ratio, 'service':service_ratio, 'ambience':ambience_ratio, 'others':others_ratio }, ignore_index=True)
# Writing to file
rest_df.to_csv("result/question2.csv")
