Analytics & dashboardsHR
Employee sentiment
Spot unhappy employees early, from the tone of their work emails
2,191work emails read for tone
See the code on GitHub (opens in a new tab)
Proof you can check:Code on GitHub
Analytics & dashboards
7 possible flight risks, spotted from tone alone
- Neutral 1,056
- Positive 971
- Negative 164
- 1 dot = 10 emails
Lubo Bali · lubobali.com
The problem
HR often learns that someone is unhappy only when they resign, and by then it's too late to help.
What I built
Python reads each email and marks it positive, neutral or negative, scores every person month by month, and flags anyone with 4 or more negative emails in 30 days.
Tech & why
Simple, explainable language analysis (TextBlob) and scikit-learn, so HR can see exactly why someone was flagged.
- Python
- Pandas
- TextBlob
- scikit-learn
The result
92% of emails were neutral or positive, so the workplace was mostly healthy, but 7 people were flagged as possible flight risks (people who may quit), even some of the most positive ones.
Similar projects
- Analytics & dashboardsWhat really drives pay and quittingTested HR beliefs with statistics: experience, not age, drives pay, and a firing rumor was false60% of pay explained by experience + age (25% by age alone)
- Analytics & dashboardsMining process qualityWhich plant settings give the purest iron, from sensor data taken every 20 seconds9.8–10.2 pulp pH where the best iron yield happens
- Analytics & dashboardsContractor performance (modeled on CVS)Which maintenance contractors cost the most and miss their deadlines18% of work orders overdue, on average
Have a problem like this? Get your free AI blueprint in about a minute.
