()
| 72 | |
| 73 | |
| 74 | def split_test(): |
| 75 | fi = open("jointed-new", "r") |
| 76 | fo = open("jointed-new-split-info", "w") |
| 77 | user_count = {} |
| 78 | for line in fi: |
| 79 | line = line.strip() |
| 80 | user = line.split("\t")[1] |
| 81 | if user not in user_count: |
| 82 | user_count[user] = 0 |
| 83 | user_count[user] += 1 |
| 84 | fi.seek(0) |
| 85 | i = 0 |
| 86 | last_user = "A26ZDKC53OP6JD" |
| 87 | for line in fi: |
| 88 | line = line.strip() |
| 89 | user = line.split("\t")[1] |
| 90 | if user == last_user: |
| 91 | if i < user_count[user] - 2: # 1 + negative samples |
| 92 | print("20180118" + "\t" + line, file=fo) |
| 93 | else: |
| 94 | print("20190119" + "\t" + line, file=fo) |
| 95 | else: |
| 96 | last_user = user |
| 97 | i = 0 |
| 98 | if i < user_count[user] - 2: |
| 99 | print("20180118" + "\t" + line, file=fo) |
| 100 | else: |
| 101 | print("20190119" + "\t" + line, file=fo) |
| 102 | i += 1 |
| 103 | |
| 104 | |
| 105 | process_meta(sys.argv[1]) |
no test coverage detected