Update app.py
Browse files
app.py
CHANGED
@@ -105,7 +105,6 @@ def dataframe_preprocessing(df_to_preprocess:pd.DataFrame):
|
|
105 |
df_to_preprocess[feature] = df_to_preprocess[feature].replace('\n', ' ', regex=True).replace('\r', '', regex=True)
|
106 |
# Renaming columns (for greater model intelligibility).
|
107 |
df_to_preprocess.rename(columns={"userverified": "user is verified",
|
108 |
-
"userurl": "user has url",
|
109 |
"userdescription": "user description",
|
110 |
"usercreated_at": "user created at",
|
111 |
"followers_count": "followers count",
|
@@ -113,8 +112,6 @@ def dataframe_preprocessing(df_to_preprocess:pd.DataFrame):
|
|
113 |
"tweet_count": "tweet count",
|
114 |
"userlocation": "user location"},
|
115 |
inplace=True)
|
116 |
-
# Making URL column binary.
|
117 |
-
df_to_preprocess['user has url'].replace({'_URL_': 'True', "": 'False'}, inplace=True)
|
118 |
# Adding some extra features.
|
119 |
df_to_preprocess['years since account created'] = df_to_preprocess['created_at'].dt.year.astype('Int64') - df_to_preprocess['user created at'].dt.year.astype('Int64')
|
120 |
df_to_preprocess['tweets per day'] = df_to_preprocess['tweet count']/((df_to_preprocess['created_at'] - df_to_preprocess['user created at']).dt.days)
|
@@ -257,7 +254,7 @@ def get_user_tweets(user_id:str, days_to_go_back:int, client:tw.Client):
|
|
257 |
start_time='{}-{}-{}T{}:00:00Z'.format(year,month,day,hour),
|
258 |
tweet_fields=['author_id','created_at','public_metrics','source'],
|
259 |
until_id=None,
|
260 |
-
user_fields=['created_at','description','location','public_metrics','
|
261 |
user_auth=False,
|
262 |
limit=500)
|
263 |
except:
|
@@ -353,7 +350,7 @@ def on_receipt_of_tweet_query(request:str,client:tw.Client):
|
|
353 |
place_fields=None,
|
354 |
poll_fields=None,
|
355 |
tweet_fields=['author_id','created_at','public_metrics','source'],
|
356 |
-
user_fields=['created_at','description','location','public_metrics','
|
357 |
user_auth=False)
|
358 |
|
359 |
# STAGE 2. PREPROCESSING OF TWEET DATA
|
|
|
105 |
df_to_preprocess[feature] = df_to_preprocess[feature].replace('\n', ' ', regex=True).replace('\r', '', regex=True)
|
106 |
# Renaming columns (for greater model intelligibility).
|
107 |
df_to_preprocess.rename(columns={"userverified": "user is verified",
|
|
|
108 |
"userdescription": "user description",
|
109 |
"usercreated_at": "user created at",
|
110 |
"followers_count": "followers count",
|
|
|
112 |
"tweet_count": "tweet count",
|
113 |
"userlocation": "user location"},
|
114 |
inplace=True)
|
|
|
|
|
115 |
# Adding some extra features.
|
116 |
df_to_preprocess['years since account created'] = df_to_preprocess['created_at'].dt.year.astype('Int64') - df_to_preprocess['user created at'].dt.year.astype('Int64')
|
117 |
df_to_preprocess['tweets per day'] = df_to_preprocess['tweet count']/((df_to_preprocess['created_at'] - df_to_preprocess['user created at']).dt.days)
|
|
|
254 |
start_time='{}-{}-{}T{}:00:00Z'.format(year,month,day,hour),
|
255 |
tweet_fields=['author_id','created_at','public_metrics','source'],
|
256 |
until_id=None,
|
257 |
+
user_fields=['created_at','description','location','public_metrics','verified'],
|
258 |
user_auth=False,
|
259 |
limit=500)
|
260 |
except:
|
|
|
350 |
place_fields=None,
|
351 |
poll_fields=None,
|
352 |
tweet_fields=['author_id','created_at','public_metrics','source'],
|
353 |
+
user_fields=['created_at','description','location','public_metrics','verified'],
|
354 |
user_auth=False)
|
355 |
|
356 |
# STAGE 2. PREPROCESSING OF TWEET DATA
|