30:T664,# fedLog ^{[code][issues]} Experimenting with in-browser Federated Logistic Regression using Box.com as the backend.

## Abstract - do we already have what we need? The profusion of consumer-facing cloud services is making commoditized high-end computational infrastructure available to epidemiological studies. Case in point, cloud middle layers like Box.com and Google Drive come with security, compliance (HIPPA) and co-location settings (GDPR) operable by an extensive API (Application Programming Interface) tier. In principle, this functionality should suffice to devise federated learning where multiple cohorts can parameterize a shared model without ever aggregating the data. To test the feasibility of such a consumer-facing cloud-hosted construct, the federated parameterization of logistic regression will be attempted here using box.com as the backend. Each participant will manage a separate data file, and all will have r/w permissions for a single shared file keeping parameter values. The libraries of two other projects will be used to orchestrate the federation: * Logistic regression using fminsearch: for example https://episphere.github.io/logist. * Authenticated operating Box APIs: https://observablehq.com/@episphere/epibox. ## Authentication Load epibox and authenticate your Box login 31:[] 2e:{"title":"fedLog [code][issues]","cells":[{"id":0,"value":"$30","pinned":false,"hidden":false,"mode":"md"},{"id":25,"value":"epibox = require(\"https://episphere.github.io/epibox/epibox.js\")\n//epibox = (await import(\"https://episphere.github.io/epibox/export.js\")).epibox","pinned":false,"hidden":false,"mode":"ojs"},{"id":30,"value":"epibox.loginObservable()","pinned":true,"hidden":false,"mode":"ojs"},{"id":34,"value":"Once logged in, check that clicking on the \"User\" button will correctly identify you as the user.\n\n## Library\nThe support library, **fedlog**, was imported from https://episphere.github.io/episphere/fedlog/fedlog.mjs. You have been using this library already - this is a reactive notebook, among other things managing authentication, [above](https://observablehq.com/@episphere/fedlog#cell-30). \n","pinned":false,"hidden":false,"mode":"md"},{"id":150,"value":"// fedlog = import(\"http://localhost:8000/fedlog/fedlog.mjs\")\nfedlog = import(\"https://epiverse.github.io/fedlog/fedlog.mjs\")","pinned":true,"hidden":false,"mode":"ojs"},{"id":148,"value":"## Your data (private)\nThis will be a csv file with your private data. It can be a private file you keep in Box, or it can be a local content you load into the memory of this machine. Either way, this data does not need to be available to other users, and it will not be in any way shared or circulated outside this browser session. You can provide your private data in one of three ways - as a private box file id, as a local file, or through copy/paste onto a text area:","pinned":false,"hidden":false,"mode":"md"},{"id":43,"value":"viewof fID = Inputs.text({ label: \"Box file ID:\" })","pinned":false,"hidden":false,"mode":"ojs"},{"id":48,"value":"viewof file = Inputs.file({ label: \"upload\" })","pinned":false,"hidden":false,"mode":"ojs"},{"id":59,"value":"viewof textarea = Inputs.textarea({ label: \"Paste\" })","pinned":false,"hidden":false,"mode":"ojs"},{"id":63,"value":"viewof buttonSetosa = Inputs.button(\"Iris Setosa sepalLength\")","pinned":false,"hidden":false,"mode":"ojs"},{"id":132,"value":"clickSetosa = {\n buttonSetosa;\n return Date();\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":67,"value":"viewof buttonNonSetosa = Inputs.button(\"Iris nonSetosa sepalLength\")","pinned":true,"hidden":false,"mode":"ojs"},{"id":88,"value":"### Demo data - aggregate\nAs a demo, we'll distribute different parts of the iris dataset between two different users (click one of the two Iris buttons above). To see the logistic regression with the full aggregated dataset see episphere.github.io/logist. ","pinned":false,"hidden":false,"mode":"md"},{"id":77,"value":"iris = (await fetch(\"https://episphere.github.io/ai/data/iris.json\")).json()","pinned":true,"hidden":false,"mode":"ojs"},{"id":86,"value":"data = iris","pinned":true,"hidden":false,"mode":"ojs"},{"id":108,"value":"### Demo data - your private data\nThe data subset you chose above by clicking in one of the two demo buttons","pinned":false,"hidden":false,"mode":"md"},{"id":110,"value":"buttonNonSetosa","pinned":true,"hidden":false,"mode":"ojs"},{"id":112,"value":"buttonSetosa.button","pinned":true,"hidden":false,"mode":"ojs"},{"id":116,"value":"lala = Date()","pinned":true,"hidden":false,"mode":"ojs"},{"id":137,"value":"iris.filter((x) => x.species == \"setosa\").map((x) => x.sepalLength)","pinned":true,"hidden":false,"mode":"ojs"}],"files":"$W31","stdlib":"1","titling":"auto"}