thanks for catching that! this was a typo, fixed now :p
Using this as example, here's how an insight like this comes about:
- user downloads data from different sources to device
- We use activitywatch (https://activitywatch.net/) to know what activities I'm doing on the computer + when it happens and auto categorize into different groups (e.g neurotech, angel investing etc)
- Exercise/high intensity training data is obtained from Oura ring (https://cloud.ouraring.com/docs/activity) - e.g `activity.high` (number of mins of high intensity training in a day), `activity.met_1min` (metabolic equivalent of activity in 1min granularity) variables
After pre-processing, we combine these features (amongst others), and the calculate correlation matrices + conditional probability to see the impact of given variables on desired outcome.
It still pretty early days and we hope to generate higher signal predictions from aggregated sources trying out different modelling techniques.
Using this as example, here's how an insight like this comes about:
- user downloads data from different sources to device
- We use activitywatch (https://activitywatch.net/) to know what activities I'm doing on the computer + when it happens and auto categorize into different groups (e.g neurotech, angel investing etc)
- Exercise/high intensity training data is obtained from Oura ring (https://cloud.ouraring.com/docs/activity) - e.g `activity.high` (number of mins of high intensity training in a day), `activity.met_1min` (metabolic equivalent of activity in 1min granularity) variables
After pre-processing, we combine these features (amongst others), and the calculate correlation matrices + conditional probability to see the impact of given variables on desired outcome.
It still pretty early days and we hope to generate higher signal predictions from aggregated sources trying out different modelling techniques.