#Issue 26 - What can phone data reveal about your wealth?
Machine Learning for mapping poverty data in developing countries
Reliable data is necessary for creating policies that affects social and economic growth. Although wealthy countries can spend money and resources for mass surveys and consensus to construct big data, the same is not possible for developing countries.
A paper titled Estimating Poverty Using Cell Phone Data: Evidence from Guatemala used Cell Phone Records(CDR) to create a model and compared the result with traditional surveys.
The method used supervised machine learning, where CDR data predicted poverty rates. It used both regression (predicting exact poverty rates) and classification (grouping poverty into categories like low, moderate, or high). Some CDR data includes-
It reveals that while CDRs can effectively predict urban and total poverty rates, their accuracy diminishes in rural areas due to lower cellular penetration and data granularity. Overall, models that categorised poverty rates into three groups were more accurate because it was easier than predicting exact poverty levels.
Certain patterns have emerged from data in how rich and poor people use their mobile phones. According to Joshua Blumenstock, author of the paper Predicting poverty and wealth from mobile phone metadata wealthier individuals tend to make more calls, while poorer people are more likely to receive them. Those who make calls during regular business hours are likely different from those who call at odd times, possibly because they are more likely to have office jobs. Additionally, poorer individuals tend to top-up their phones in small, frequent amounts.
Some other papers which have used phone metadata to map data for good uses-
Machine learning and phone data can improve targeting of humanitarian aid
Behavior revealed in mobile phone usage predicts credit repayment
The use of mobile phone data to inform analysis of COVID-19 pandemic epidemiology
Links Around The Web
12 Ways To Tell If You’re A Difficult Engineer (And What To Do About It)
Using GPT-4 to generate 100 words consumes up to 3 bottles of water
Books I am Reading
I am currently reading Making of a Manager by Julie Zhuo, which one of my colleagues gifted me on my birthday. The book covers various management aspects like leading teams, hiring, meetings, and feedback. Much of the advice feels obvious and can be learned on the job, but overall not a bad read.
Some other books I have read over the last few weeks and some favorite quotes from them-
In the Miso Soup by Ryu Murakami
I once heard a psychiatrist type say on TV that people need to feel they’re of some value to go on living, and I think there’s something to that. It wouldn’t be easy to keep going if you thought you were of no use to anyone.
How Big Things Get Done by Bent Flyvbjerg, Dan Gardner
The common denominator of any project is that people are making the decisions about it. And wherever there are people, there are psychology and power.
Planning requires thinking—and creative, critical, careful thinking is slow
Inspired: Building products customer love by Marty Cagan
The honest truth is that the product manager needs to be among the strongest talent in the company. If the product manager doesn't have the technology sophistication, doesn't have the business savvy, doesn't have the credibility with the key executives, doesn't have the deep customer knowledge, doesn't have the passion for the product, or doesn't have the respect of their product team, then it's a sure recipe for failure.
That’s it folks. Thanks for reading.