Table 30 Top five attributes selected and their respective information gain value for both Android and Swing training sets on the three settings for the tests
From: Redocumenting APIs with crowd knowledge: a coverage analysis based on question types
Setting | #Selected | Top five attributes | Info gain value |
|---|---|---|---|
1B | 10/12 | QUESTION_CODE_SIZE | 0.377347749 |
DEBUG_KEYWORDS | 0.330994105 | ||
QUESTION_HAS_CODE | 0.305653075 | ||
QUESTION_BODY_SIZE | 0.27711226 | ||
HOW_KEYWORDS | 0.178311939 | ||
2B | 32/153 | QUESTION_CODE_SIZE | 0.377347749 |
QUESTION_HAS_CODE | 0.305653075 | ||
QUESTION_BODY_SIZE | 0.27711226 | ||
howto_KEYWORD | 0.1949863 | ||
error_KEYWORD | 0.09552362 | ||
3B | 10/12 | QUESTION_CODE_SIZE | 0.377347749 |
DEBUG_KEYWORDS | 0.321029275 | ||
QUESTION_HAS_CODE | 0.305653075 | ||
QUESTION_BODY_SIZE | 0.27711226 | ||
HOW_KEYWORDS | 0.194580371 |