Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

< Back to Article

Table 1.

Emoji usage on GitHub by post type.

Over 5% of posts on GitHub contain at least one emoji. Pull request comments and commit comments have higher proportions of emoji posts.

More »

Table 1 Expand

Fig 1.

Emoji usage varies in repositories that have different primary programming languages.

Dot: a language; size: number of repositories; color: proportion of emoji posts. JavaScript developers use a larger variety of emojis and a larger proportion of their posts contain emojis.

More »

Fig 1 Expand

Fig 2.

Working status of a developer is related to the number of emoji posts.

(a): Avg. daily working hours. (b): Prop. push events.

More »

Fig 2 Expand

Table 2.

OLS regressions of emoji usage on working status.

Multiple working status measures are significantly related to emoji usage.

More »

Table 2 Expand

Fig 3.

Accuracy of dropout predictions for emoji users is above 0.7 when including all activity levels and as high as 0.75 for those who work longer hours per working day.

Baseline is 0.5 for all activity levels and is omitted. (a): Avg. daily working hours. (b): #working days.

More »

Fig 3 Expand

Fig 4.

AUC of dropout predictions for emoji users is above 0.78 when including all activity levels and as high as 0.82 for those who work longer hours per working day.

(a): Avg. daily working hours. (b): #working days.

More »

Fig 4 Expand

Fig 5.

Importance of top features (10% active users by avg. working hours).

Proportion of emoji posts, emotional scores, and affection-related emojis are important predictors. Bars: importance scores from GBDT; signed values: coefficients from Logistic Regression.

More »

Fig 5 Expand

Fig 6.

Relation between selected emoji usage features and dropout risk (10% active users by avg. working hours).

Blue dots: bins of emoji users; orange diamond: non-emoji users. (a): #Emoji posts (log). (b): Prop. emoji posts. (c): Emoji entropy. (d): Score of Anger emotion.

More »

Fig 6 Expand

Fig 7.

Emotion expressed via emojis in 2019 is related to the number of emoji posts in 2018.

(a): Positive emotion. (b): Negative emotion.

More »

Fig 7 Expand

Table 3.

Positive and negative emotions expressed via emojis in 2018 and 2019.

In general, the population becomes more positive and less negative in 2019. Those who used any emoji in 2018 are more positive (p < 0.001, Welch two-sample t-test) and less negative (p < 0.001, Welch two-sample t-test) in 2019 than those who didn’t use an emoji in 2018.

More »

Table 3 Expand

Table 4.

Difference of responses between issues with and without emoji.

For issues with at least one comment, the time to receive the first comment for emoji issues is significantly shorter than non-emoji issues (p < 0.001, Welch two-sample t-test). Emoji issues get significantly more comments containing at least one emoji (p < 0.001, Welch two-sample t-test).

More »

Table 4 Expand