How to Do Data for Good
summary
DataKind UK pairs volunteer data people with charities via monthly data dives and longer data cores, then applies practical methods: a regression to impute missing FOI data produces a UK youth homelessness estimate of ~138,000 (≈12× the official figure); NLP topic modelling on 10,000 grant texts surfaces chains like domestic violence → child mental‑health and links between debt and bedwetting; a graph database + red‑flag dashboard surfaces circular ownership and 400 politician officers, prompting Companies House fixes; a random forest dependency score predicts repeat food‑bank use and becomes integral after nine months.
AI-written summary · about
chapters (37)
- 0:00Intro
- 0:28DataKind UK
- 0:42Mission
- 2:06Programs
- 3:40Understanding the needs and demand
- 8:27Buttle UK
- 11:35Campaigning
- 12:23Global Witness
- 14:11Trump
- 17:01Circular structures
- 18:06Red flags
- 19:46Data limitations
- 20:40Global Witness analysis
- 21:27crackdowns
- 22:01money laundering
- 23:40evaluation
- 24:15Welcome Centre
- 24:37Welcome Centre referrals
- 28:54How data is collected
- 29:18Openended flexible support
- 30:49Future usage
- 32:38Feedback from charity
- 35:27Ethics in data
- 36:56What if
- 37:41Human verification
- 41:12Dog
- 41:47Questions
- 42:03Nonprofit vs profit sector
- 43:38Remote work
- 44:34DataKind in Hungary
- 45:07How many applicants get selected
- 45:48How do you decide which project to work on
- 46:57Where do you publish your work
- 47:33Currently running projects
- 48:25Parttime volunteers
- 49:22Helping a charity
- 50:31Identifying causes of homelessness
related talks
47mData for good: Lessons from the frontline
17mKeynote
24mData science, apps and civic responsibility
12mThe Analytical Imperative
41mUsing web data for a better tomorrow
18mCalling All Data Scientists for the Common Good in Cities
11mConnecting Researchers to Patient and Person Generated Data
13mData Nerding in Public Health