import pandas as pd
# Load in all the data
chat_df = pd.read_csv("chat.csv")
match_df = pd.read_csv("match.csv")
match_df.set_index('match_id', inplace=True)
# Using a bag of words model to rank each phrase
with open('positive_words.txt') as pos_words_file:
positive_words = set(pos_words_file.read().split('\n')[:-1])
with open('negative_words.txt') as neg_words_file:
negative_words = set(neg_words_file.read().split('\n')[:-1])# Analyse sentences and retrieve scores
def score_chat(row):
score = 0
for word in str(row['key']).split(" "):
if word in positive_words:
score += 1
if word in negative_words:
score -= 1
return score
scores = chat_df.apply(score_chat, axis=1)
chat_df['score'] = scores
chat_df.drop(['key', 'slot', 'time', 'unit'], axis=1, inplace=True)
chat_df.to_csv('processed_chat.csv')score_df = pd.read_csv('processed_chat.csv')# Load clusters into score dataframe
def region_from_match_id(row):
return match_df.iloc[row['match_id']]['cluster']
score_df['cluster'] = score_df.apply(region_from_match_id, axis=1)
score_df.to_csv('processed_chat.csv')from collections import defaultdict
import json
# Tally scores based on region
with open('region.json') as regions_file:
regions = json.load(regions_file)
tally = defaultdict(int)
for region_code, region in regions.items():
region_chats = score_df[score_df['cluster'] == int(region_code)]
region_score = region_chats['score'].sum()
tally[region] += region_score# Plot data on map
print("======= Toxicity Scale =======")
for n, country in enumerate(sorted(tally, key=lambda x: tally[x]), start=1):
print(f"{n}. {country} - {tally[country]}")======= Toxicity Scale =======
1. EUROPE - -12318
2. SINGAPORE - -9713
3. US EAST - -8739
4. US WEST - -4387
5. AUSTRALIA - -4324
6. AUSTRIA - -1376
7. STOCKHOLM - -239
8. JAPAN - -174
9. DUBAI - -98
10. PW TELECOM SHANGHAI - -10
11. CHILE - -1
12. SOUTHAFRICA - 0
13. PW TELECOM GUANGDONG - 0
14. PW TELECOM WUHAN - 0
15. PW UNICOM - 0
16. PW UNICOM TIANJIN - 0
17. PERU - 0
18. INDIA - 0
19. PW TELECOM ZHEJIANG - 2
20. BRAZIL - 69