from database import Firebase
firebase = Firebase()
count = firebase.get_program_count()
admission_list = []
param = 0.10
for pid in range(0, count):
res = firebase.get_program_admission_rate(pid)
if res >= param:
admission_list.append(pid)
print admission_list
from database import Firebase
firebase = Firebase()
count = firebase.get_program_count()
city_list = []
param = "Seattle"
for pid in range(0, count):
res = firebase.get_program_location(pid)
if res['city'] == param:
city_list.append(pid)
print city_list
def __init__(self):
# filtered_list=[]
self.firebase = Firebase()
self.count = self.firebase.get_program_count()
from database import Firebase firebase = Firebase() count = firebase.get_program_count() verbal_list=[] param=150 for pid in range(0, count): res=firebase.get_program_acad(pid) for each in res: if each=='gre': temp=res[each] val=temp['verbal'] if val<=param: verbal_list.append(pid) print verbal_list
from database import Firebase
firebase = Firebase()
count = firebase.get_program_count()
fees_in_list = []
param = 50000
for pid in range(0, count):
res = firebase.get_program_fees(pid)
if res['in_state'] <= param:
fees_in_list.append(pid)
print fees_in_list
class Filters(object):
def __init__(self):
# filtered_list=[]
self.firebase = Firebase()
self.count = self.firebase.get_program_count()
def filter_rank(self, param):
param = int(param)
rank_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_rank(pid)
if res['cwur'] <= param:
rank_list.append(pid)
if res['usnews'] <= param:
rank_list.append(pid)
if res['forbes'] <= param:
rank_list.append(pid)
if res['times'] <= param:
rank_list.append(pid)
rank_final = set(rank_list)
return list(rank_final)
def filter_rank_absolute(self, param):
param = int(param)
rank_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_rank(pid)
if res['cwur'] <= param and res['usnews'] <= param and res[
'forbes'] <= param and res['times'] <= param:
rank_list.append(pid)
return rank_list
def filter_rank_overall(self, param):
param = int(param)
overall_list = []
for pid in range(0, param + 1):
res = self.firebase.get_program_rank(pid)
overall_list.append(pid)
return overall_list
def filter_rank_usnews(self, param):
usnews_list = []
param = int(param)
for pid in range(0, self.count):
res = self.firebase.get_program_rank(pid)
if res['usnews'] <= param:
usnews_list.append(pid)
return usnews_list
def filter_rank_cwur(self, param):
cwur_list = []
param = int(param)
for pid in range(0, self.count):
res = self.firebase.get_program_rank(pid)
if res['cwur'] <= param:
cwur_list.append(pid)
return cwur_list
def filter_rank_forbes(self, param):
forbes_list = []
param = int(param)
for pid in range(0, self.count):
res = self.firebase.get_program_rank(pid)
if res['forbes'] <= param:
forbes_list.append(pid)
return forbes_list
def filter_rank_times(self, param):
times_list = []
param = int(param)
for pid in range(0, self.count):
res = self.firebase.get_program_rank(pid)
if res['times'] <= param:
times_list.append(pid)
return times_list
# 2. Location
#{u'city': u'Philadelphia', u'region_id': 2, u'zip': 19104, u'lon': -75.193618, u'state': u'PA', u'lat': 39.951002, u'region_name': u'Mid East'}
def filter_location_state(self, param):
state_list = []
for each in param:
for pid in range(0, self.count):
res = self.firebase.get_program_location(pid)
if res['state'] == each:
state_list.append(pid)
return state_list
# return list(set(state_list)&set(filtered_list));
def filter_location_city(self, param):
city_list = []
for each in param:
for pid in range(0, self.count):
res = self.firebase.get_program_location(pid)
if res['city'] == param:
city_list.append(pid)
return city_list
# return list(set(city_list)&set(filtered_list));
def filter_location_zip(self, param):
zip_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_location(pid)
if res['zip'] == param:
zip_list.append(pid)
return zip_list
# return list(set(zip_list)&set(filtered_list));
def filter_location_region(self, param):
region_list = []
region_dict = {
"north_east": ["CT", "ME", "MA", "NH", "RI", "VT"],
"south": [
"DE", "DC", "MD", "AL", "AR", "FL", "GA", "KY", "LA", "MS",
"NC", "SC", "TN", "VA", "WV"
],
"mid_west": [
"IL", "IN", "MI", "OH", "WI", "IA", "KS", "MN", "MO", "NE",
"ND", "SD"
],
"west": [
"AZ", "NM", "CO", "ID", "MT", "UT", "WY", "AK", "CA", "HI",
"NV", "OR", "WA"
]
}
for each in param:
states_list = region_dict[each]
for every_state in states_list:
for pid in range(0, self.count):
res = self.firebase.get_program_location(pid)
if res['state'] == every_state:
region_list.append(pid)
return region_list
# return list(set(resion_list)&set(filtered_list));
# 3. Fees
#{u'out_of_state': 63000, u'in_state': 42690}
def filter_fees_in_state(self, param):
fees_in_list = []
param = int(param)
for pid in range(0, self.count):
res = self.firebase.get_program_fees(pid)
if res['in_state'] <= param:
fees_in_list.append(pid)
return fees_in_list
# return list(set(fees_in_list)&set(filtered_list));
def filter_fees_out_state(self, param):
fees_out_list = []
param = int(param)
for pid in range(0, self.count):
res = self.firebase.get_program_fees(pid)
if int(res['out_of_state']) <= int(param):
fees_out_list.append(pid)
return fees_out_list
# return list(set(fees_out_list)&set(filtered_list));6
def filter_budget(self, budget):
budget_list = []
budget = int(budget)
for pid in range(0, self.count):
res1 = self.firebase.get_program_details(pid)
duration = float(res1['length'])
res2 = self.firebase.get_program_fees(pid)
total_out_of_state_fees = float(res2['out_of_state'])
if duration > 12:
yearly_out_of_state = float(
12.0 / duration) * total_out_of_state_fees
else:
yearly_out_of_state = total_out_of_state_fees
res3 = self.firebase.get_program_living(pid)
total_living = float(res3['overall'])
total_cost_per_year = yearly_out_of_state + total_living
if total_cost_per_year < budget:
budget_list.append(pid)
return budget_list
# 4. Score
#{u'gre': {u'quant': 167, u'verbal': 159}, u'gpa': 3}
def filter_gpa(self, param):
gpa_list = []
param = float(param)
for pid in range(0, self.count):
res = self.firebase.get_program_acad(pid)
if res['gpa'] <= param:
gpa_list.append(pid)
return gpa_list
# return list(set(fees_list)&set(filtered_list));
def filter_gre_verbal(self, param):
verbal_list = []
param = int(param)
for pid in range(0, self.count):
res = self.firebase.get_program_acad(pid)
for each in res:
if each == 'gre':
temp = res[each]
val = temp['verbal']
if val <= param:
verbal_list.append(pid)
return verbal_list
# return list(set(fees_list)&set(filtered_list));
def filter_gre_quant(self, param):
quant_list = []
param = int(param)
for pid in range(0, self.count):
res = self.firebase.get_program_acad(pid)
for each in res:
if each == 'gre':
temp = res[each]
val = temp['quant']
if val <= param:
quant_list.append(pid)
return quant_list
# return list(set(quant_list)&set(filtered_list));
# 5. Living
#{u'boarding': 14601, u'books': 1425, u'overall': 19351, u'other': 3325}
def filter_boarding(self, param):
boarding_list = []
param = int(param)
for pid in range(0, self.count):
res = self.firebase.get_program_living(pid)
if res['boarding'] <= param:
boarding_list.append(pid)
return boarding_list
# return list(set(boarding_list)&set(filtered_list));
def filter_books(self, param):
books_list = []
param = int(param)
for pid in range(0, self.count):
res = self.firebase.get_program_living(pid)
if res['books'] <= param:
books_list.append(pid)
return books_list
# return list(set(books_list)&set(filtered_list));
def filter_overall_expenses(self, param):
overall_list = []
param = int(param)
for pid in range(0, self.count):
res = self.firebase.get_program_living(pid)
if res['overall'] <= param:
overall_list.append(pid)
return overall_list
# return list(set(overall_list)&set(filtered_list));
# 6. Admission Rate
# Filter by Admission Rate
def filter_admission_rate(self, param):
admission_list = []
param = float(param)
for pid in range(0, self.count):
res = self.firebase.get_program_admission_rate(pid)
if res >= param:
admission_list.append(pid)
return admission_list
class Recommend(object):
"""Class for Selecting Programs Matching Filters."""
def __init__(self,
overall_rank=None,
rank=None,
usnews=None,
cwur=None,
forbes=None,
times=None,
state=None,
city=None,
zipcode=None,
region=None,
budget=None,
location=None,
in_state=None,
out_of_state=None,
gpa=None,
verbal=None,
quant=None,
boarding=None,
books=None,
overall_expenses=None,
admission_rate=None,
aoi=None):
"""Initializing Filter Class."""
self.rank = rank
self.overall_rank = overall_rank
self.usnews = usnews
self.cwur = cwur
self.forbes = forbes
self.times = times
self.state = state
self.city = city
self.zipcode = zipcode
self.region = region
self.location = location
self.in_state = in_state
self.out_of_state = out_of_state
self.gpa = gpa
self.verbal = verbal
self.quant = quant
self.budget = budget
self.boarding = boarding
self.books = books
self.overall_expenses = overall_expenses
self.admission_rate = admission_rate
self.aoi = aoi
# List of programs matching atleast one of the requested criteria
self.unique_programs = None
# List of programs meeting all requested criteria
self.common_programs = []
# Verbose program dict with all matching info
self.matches = None
# creating the object of Filters class
self.filters = Filters()
# Program Count
self.count = self.filters.count
self.firebase = Firebase()
def filter_programs(self):
"""Filter the programs."""
# Deploy Filter Selection
specializations, user_selected_criteria = self.run_selected_filters()
# Getting all the unique programs
self.unique_programs = list(self.matches.keys())
# Getting all common program intersection
common = set.intersection(*map(set, self.common_programs))
# Validating we atleast have one program matching one criteria
# Fall back to all programs if no match is found
if len(self.unique_programs) < 1:
# Setting unique programs to all programs
self.unique_programs = []
self.unique_programs.extend(range(0, self.count - 1))
# Setting all matches to zero
self.matches = []
self.matches = dict.fromkeys(self.unique_programs, 1)
# Setting all common programs to all programs
common = []
common = common.extend(range(0, self.count - 1))
return (common, self.unique_programs, self.matches, specializations,
user_selected_criteria)
def update_results(self, matching_programs):
"""Helper method to update matching programs results."""
self.common_programs.append(matching_programs)
for program in matching_programs:
# Update the verbose program dict of results
if self.matches is None:
self.matches = {}
self.matches[program] = 1
elif program in self.matches:
self.matches[program] += 1
else:
self.matches[program] = 1
def run_selected_filters(self):
specializations = []
user_selected_criteria = []
#Based on the UI Options
if self.rank is not None:
matching_programs = self.filters.filter_rank_usnews(self.rank)
self.update_results(matching_programs)
matching_programs = self.filters.filter_rank_cwur(self.rank)
self.update_results(matching_programs)
matching_programs = self.filters.filter_rank_forbes(self.rank)
self.update_results(matching_programs)
matching_programs = self.filters.filter_rank_times(self.rank)
self.update_results(matching_programs)
rank_list = ["usnews", "cwur", "forbes", "times"]
user_selected_criteria.extend(rank_list)
if self.budget is not None:
matching_programs = self.filters.filter_fees_out_state(self.budget)
self.update_results(matching_programs)
matching_programs = self.filters.filter_fees_in_state(self.budget)
self.update_results(matching_programs)
matching_programs = self.filters.filter_boarding(self.budget)
self.update_results(matching_programs)
matching_programs = self.filters.filter_books(self.budget)
self.update_results(matching_programs)
user_selected_criteria.append("in_state")
user_selected_criteria.append("out_of_state")
user_selected_criteria.append("boarding")
user_selected_criteria.append("books")
if self.aoi is not None:
specializations.extend(self.aoi)
if self.location is not None:
matching_programs = self.filters.filter_location_region(
self.location)
self.update_results(matching_programs)
"""Individual filters."""
if self.overall_rank is not None:
matching_programs = self.filters.filter_rank_overall(
self.overall_rank)
self.update_results(matching_programs)
if self.usnews is not None:
matching_programs = self.filters.filter_rank_usnews(self.usnews)
self.update_results(matching_programs)
if self.cwur is not None:
matching_programs = self.filters.filter_rank_cwur(self.cwur)
self.update_results(matching_programs)
if self.forbes is not None:
matching_programs = self.filters.filter_rank_forbes(self.forbes)
self.update_results(matching_programs)
if self.times is not None:
matching_programs = self.filters.filter_rank_times(self.times)
self.update_results(matching_programs)
if self.state is not None:
matching_programs = self.filters.filter_location_state(self.state)
self.update_results(matching_programs)
if self.city is not None:
matching_programs = self.filters.filter_location_city(self.city)
self.update_results(matching_programs)
if self.zipcode is not None:
matching_programs = self.filters.filter_location_zip(self.zipcode)
self.update_results(matching_programs)
if self.in_state is not None:
matching_programs = self.filters.filter_fees_in_state(
self.in_state)
self.update_results(matching_programs)
if self.out_of_state is not None:
matching_programs = self.filters.filter_fees_out_state(
self.out_of_state)
self.update_results(matching_programs)
if self.gpa is not None:
matching_programs = self.filters.filter_gpa(self.gpa)
self.update_results(matching_programs)
if self.verbal is not None:
matching_programs = self.filters.filter_gre_verbal(self.verbal)
self.update_results(matching_programs)
if self.quant is not None:
matching_programs = self.filters.filter_gre_quant(self.quant)
self.update_results(matching_programs)
if self.boarding is not None:
matching_programs = self.filters.filter_boarding(self.boarding)
self.update_results(matching_programs)
if self.books is not None:
matching_programs = self.filters.filter_books(self.books)
self.update_results(matching_programs)
if self.overall_expenses is not None:
matching_programs = self.filters.filter_overall_expenses(
self.overall_expenses)
self.update_results(matching_programs)
if self.admission_rate is not None:
matching_programs = self.filters.filter_admission_rate(
self.admission_rate)
self.update_results(matching_programs)
return specializations, user_selected_criteria
def get_user_criteria(self):
if self.rank is not None:
self.user_selected_criteria.extend("usnews")
self.user_selected_criteria.extend("forbes")
self.user_selected_criteria.extend("times")
self.user_selected_criteria.extend("cwur")
if self.in_state is not None:
self.user_selected_criteria.extend("in_state")
self.user_selected_criteria.extend("out_of_state")
self.user_selected_criteria.extend("boarding")
self.user_selected_criteria.extend("books")
self.user_selected_criteria.extend("admission_rate")
if self.gpa is not None:
self.user_selected_criteria.extend("gpa")
if self.verbal is not None:
self.user_selected_criteria.extend("verbal")
if self.quant is not None:
self.user_selected_criteria.extend("quant")
if self.aoi is not None:
self.specializations_list.extend(aoi)
else:
self.specializations_list = [
'information_assurance_cyber_security',
'business_intelligence', 'computer_networks',
'web_application_development', 'library_science',
'management_consulting', 'human_center_design_engineering',
'information_architecture', 'software_engineering',
'data_science_analytics', 'distributed_systems'
]
return (self.user_selected_criteria, self.specializations_list)
#Getting the json list after the filtering
def get_filtered_json(self, program_list):
json_list = []
for each in program_list:
result = self.firebase.get_detailed_program(each)
json_list.append(result)
return json_list
def construct_dataframe(self, json_list, program_list):
df = pd.io.json.json_normalize(json_list)
df.columns = df.columns.map(lambda x: x.split(".")[-1])
c_list = list(df.columns.values)
col_list = []
for each in c_list:
each = each.encode("utf-8")
col_list.append(each)
df = df.loc[:, [
'gpa', 'quant', 'verbal', 'admission_rate',
'business_intelligence', 'computer_networks',
'data_science_analytics', 'distributed_systems',
'human_center_design_engineering', 'information_architecture',
'information_assurance_cyber_security', 'library_science',
'management_consulting', 'software_engineering',
'web_application_development', 'in_state', 'out_of_state',
'boarding', 'books', 'other', 'length', 'cwur', 'forbes', 'times',
'usnews'
]]
df["program_id"] = program_list
return df
def recommend_programs(self):
set_list, program_list, program_dict, specializations, user_selected_criteria = self.filter_programs(
)
json_list = self.get_filtered_json(program_list)
df = self.construct_dataframe(json_list, program_list)
#user_selected_criteria=["forbes","out_of_state"]
#specializations_list=['information_assurance_cyber_security',
# 'business_intelligence',
# 'computer_networks']
#user_selected_criteria, specializations_list=self.get_user_criteria()
match = MatchingAlgo(df, program_list, specializations,
user_selected_criteria)
recommendations = match.rank_programs()
return recommendations
def __init__(self,
overall_rank=None,
rank=None,
usnews=None,
cwur=None,
forbes=None,
times=None,
state=None,
city=None,
zipcode=None,
region=None,
budget=None,
location=None,
in_state=None,
out_of_state=None,
gpa=None,
verbal=None,
quant=None,
boarding=None,
books=None,
overall_expenses=None,
admission_rate=None,
aoi=None):
"""Initializing Filter Class."""
self.rank = rank
self.overall_rank = overall_rank
self.usnews = usnews
self.cwur = cwur
self.forbes = forbes
self.times = times
self.state = state
self.city = city
self.zipcode = zipcode
self.region = region
self.location = location
self.in_state = in_state
self.out_of_state = out_of_state
self.gpa = gpa
self.verbal = verbal
self.quant = quant
self.budget = budget
self.boarding = boarding
self.books = books
self.overall_expenses = overall_expenses
self.admission_rate = admission_rate
self.aoi = aoi
# List of programs matching atleast one of the requested criteria
self.unique_programs = None
# List of programs meeting all requested criteria
self.common_programs = []
# Verbose program dict with all matching info
self.matches = None
# creating the object of Filters class
self.filters = Filters()
# Program Count
self.count = self.filters.count
self.firebase = Firebase()
from database import Firebase firebase = Firebase() count = firebase.get_program_count() overall_list=[] param=10 ''' Overall rank is the sum of all ranks. Hence, it does not start from 1 and ends at 59. Thus, to get to 10 programs, we need to simply iterate over the top 10 program id ''' for pid in range(0, param+1): res=firebase.get_program_rank(pid) overall_list.append(pid) print overall_list
from database import Firebase
firebase = Firebase()
count = firebase.get_program_count()
overall_list = []
param = 30000
for pid in range(0, count):
res = firebase.get_program_living(pid)
if res['overall'] <= param:
overall_list.append(pid)
print overall_list
class Filters(object):
def __init__(self):
# filtered_list=[]
self.firebase = Firebase()
self.count = self.firebase.get_program_count()
def filter_rank_overall(self, param):
overall_list = []
for pid in range(0, param + 1):
res = self.firebase.get_program_rank(pid)
overall_list.append(pid)
return overall_list
def filter_rank_usnews(self, param):
usnews_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_rank(pid)
if res['usnews'] <= param:
usnews_list.append(pid)
return usnews_list
def filter_rank_cwur(self, param):
cwur_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_rank(pid)
if res['cwur'] <= param:
cwur_list.append(pid)
return cwur_list
def filter_rank_forbes(self, param):
forbes_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_rank(pid)
if res['forbes'] <= param:
forbes_list.append(pid)
return forbes_list
def filter_rank_times(self, param):
times_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_rank(pid)
if res['times'] <= param:
times_list.append(pid)
return times_list
# 2. Location
#{u'city': u'Philadelphia', u'region_id': 2, u'zip': 19104, u'lon': -75.193618, u'state': u'PA', u'lat': 39.951002, u'region_name': u'Mid East'}
def filter_location_state(self, param):
state_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_location(pid)
if res['state'] == param:
state_list.append(pid)
return state_list
# return list(set(state_list)&set(filtered_list));
def filter_location_city(self, param):
city_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_location(pid)
if res['city'] == param:
city_list.append(pid)
return city_list
# return list(set(city_list)&set(filtered_list));
def filter_location_zip(self, param):
zip_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_location(pid)
if res['zip'] == param:
zip_list.append(pid)
return zip_list
# return list(set(zip_list)&set(filtered_list));
def filter_location_region(self, param):
region_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_location(pid)
if res['region_name'] == param:
region_list.append(pid)
return region_list
# return list(set(resion_list)&set(filtered_list));
# 3. Fees
#{u'out_of_state': 63000, u'in_state': 42690}
def filter_fees_in_state(self, param):
fees_in_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_fees(pid)
if res['in_state'] <= param:
fees_in_list.append(pid)
return fees_in_list
# return list(set(fees_in_list)&set(filtered_list));
def filter_fees_out_state(self, param):
fees_out_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_fees(pid)
if int(res['out_of_state']) <= int(param):
fees_out_list.append(pid)
return fees_out_list
# return list(set(fees_out_list)&set(filtered_list));
# 4. Score
#{u'gre': {u'quant': 167, u'verbal': 159}, u'gpa': 3}
def filter_gpa(self, param):
gpa_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_acad(pid)
if res['gpa'] <= param:
gpa_list.append(pid)
return gpa_list
# return list(set(fees_list)&set(filtered_list));
def filter_gre_verbal(self, param):
verbal_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_acad(pid)
for each in res:
if each == 'gre':
temp = res[each]
val = temp['verbal']
if val <= param:
verbal_list.append(pid)
return verbal_list
# return list(set(fees_list)&set(filtered_list));
def filter_gre_quant(self, param):
quant_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_acad(pid)
for each in res:
if each == 'gre':
temp = res[each]
val = temp['quant']
if val <= param:
quant_list.append(pid)
return quant_list
# return list(set(quant_list)&set(filtered_list));
# 5. Living
#{u'boarding': 14601, u'books': 1425, u'overall': 19351, u'other': 3325}
def filter_boarding(self, param):
boarding_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_living(pid)
if res['boarding'] <= param:
boarding_list.append(pid)
return boarding_list
# return list(set(boarding_list)&set(filtered_list));
def filter_books(self, param):
books_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_living(pid)
if res['books'] <= param:
books_list.append(pid)
return books_list
# return list(set(books_list)&set(filtered_list));
def filter_overall_expenses(self, param):
overall_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_living(pid)
if res['overall'] <= param:
overall_list.append(pid)
return overall_list
# return list(set(overall_list)&set(filtered_list));
# 6. Admission Rate
# Filter by Admission Rate
def filter_admission_rate(self, param):
admission_list = []
for pid in range(0, self.count):
res = self.firebase.get_program_admission_rate(pid)
if res >= param:
admission_list.append(pid)
return admission_list
# return list(set(admission_list)&set(filtered_list));
'''
#!/usr/bin/env python
"""data.gradscout.co Flask REST API v1"""
from flask import Flask, jsonify, abort, make_response, request, url_for
from database import Firebase
app = Flask(__name__)
firebase = Firebase()
@app.route('/api/programs/<int:program_id>', methods=['GET'])
def get_detailed_program(program_id):
"""Get the Detailed Program Details for a given Program ID."""
return jsonify(firebase.get_detailed_program(program_id))
@app.route('/api/programs/', methods=['GET'])
def get_all_programs():
"""Get complete data for quicker cached access.
Returns:
programs (json): JSON Data of Complete Program Details
"""
return jsonify(firebase.data)
@app.route('/api/programs/count/', methods=['GET'])
def get_program_count():
"""Get the total count of programs available in the database.
Returns:
"""Module for cleaning courses and extracting metadata."""
import re
import json
import math
import pprint
from nltk import pos_tag
from database import Firebase
from nltk.corpus import stopwords
from nltk.tokenize import word_tokenize
from nltk.stem import PorterStemmer, WordNetLemmatizer
from sklearn.feature_extraction.text import CountVectorizer
stemmer = PorterStemmer()
lemmatiser = WordNetLemmatizer()
firebase = Firebase()
STOPWORDS = set(stopwords.words('english'))
class GetData(object):
"""Get the Input Data."""
def __init__(self):
"""Initializing with the data files."""
cur_data_file = ("/Users/bagursreenivasamurth/Dev/gradscout/"
"data/data_for_database/firebase_export.json")
course_data_file = (
"/Users/bagursreenivasamurth/Dev/gradscout/data/curriculum/cur_data.json")
self.cur_json_data = self._read_json_data(cur_data_file)