{"id":23255750,"url":"https://github.com/aorosoeon/linkedin_statistical_inference","last_synced_at":"2026-03-10T13:04:26.614Z","repository":{"id":261318363,"uuid":"848958044","full_name":"aorosoeon/linkedin_statistical_inference","owner":"aorosoeon","description":"Linkedin automation through Selenium to conduct an A/B test","archived":false,"fork":false,"pushed_at":"2024-12-14T17:55:24.000Z","size":646,"stargazers_count":5,"open_issues_count":0,"forks_count":0,"subscribers_count":1,"default_branch":"main","last_synced_at":"2025-04-06T03:44:34.943Z","etag":null,"topics":["ab-testing","gspread","linkedin","orchestration","python","selenium","shell"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/aorosoeon.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2024-08-28T18:04:05.000Z","updated_at":"2025-01-03T15:24:05.000Z","dependencies_parsed_at":"2024-11-05T22:37:46.920Z","dependency_job_id":"154148dd-195c-4f04-907a-6455900e4693","html_url":"https://github.com/aorosoeon/linkedin_statistical_inference","commit_stats":null,"previous_names":["aorosoeon/linkedin_statistical_inference"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/aorosoeon/linkedin_statistical_inference","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aorosoeon%2Flinkedin_statistical_inference","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aorosoeon%2Flinkedin_statistical_inference/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aorosoeon%2Flinkedin_statistical_inference/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aorosoeon%2Flinkedin_statistical_inference/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/aorosoeon","download_url":"https://codeload.github.com/aorosoeon/linkedin_statistical_inference/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aorosoeon%2Flinkedin_statistical_inference/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":30334412,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-03-10T12:41:07.687Z","status":"ssl_error","status_checked_at":"2026-03-10T12:41:06.728Z","response_time":106,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ab-testing","gspread","linkedin","orchestration","python","selenium","shell"],"created_at":"2024-12-19T11:31:06.776Z","updated_at":"2026-03-10T13:04:26.608Z","avatar_url":"https://github.com/aorosoeon.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# linkedin_statistical_inference\n\nIs it better to send LinkedIn connection requests with a personalized message or leave them blank? That’s the question I try to answer in this repo.\n\nTo prove that one approach has a higher acceptance rate, we need to conduct an A/B test. The main idea of an A/B test is randomization, so the decision of who gets a blank request and who gets a message should be completely random. If I try to do this manually, I will introduce my bias into the test, and it wouldn’t be within the rules of A/B testing.\n\nThe architecture below automates this process to eliminate my bias. Here is how it works:\n- A Cron job on a Raspberry Pi activates orchestrator.sh, and any error that occurs after that goes to cron.log.\n- An arrow under DAILY PROCESS represents a randomized time delay between actions, so after Cron job activation, there is a delay before main.py.\n- main.py opens the Google spreadsheet with LinkedIn links and a Chromium browser, then it opens profiles, pulls needed info, and sends them requests (with random assignment of who gets a message and who gets a blank request).\n- To be on the safe side, counter updates happen after some time in a separate file to ensure that each day the bot works with new profiles.\n- LinkedIn is tricky, so checking_invites.py verifies that these people are added and logs this in a spreadsheet.\n- checking_accepts.py pulls people who accepted my invite so that we can conduct an A/B test afterward.\n- LinkedIn doesn’t allow more than 700 sent requests, so removing_old_invite_requests.py keeps it under 600.\n\n\u003cimg src=\"imgs/architecture.png\"\u003e\n\nBelow is what the spreadsheet looks like (private data is hidden, of course):\n\n\u003cimg src=\"imgs/spreadsheet.png\"\u003e\n\nThe A/B test was recently completed with the following stats:\n\n\u003cimg src=\"imgs/results.png\" width=\"50%\" height=\"auto\"\u003e\n\nI will be releasing a Jupyter Notebook with statistical tests soon, but a preliminary analysis confirms the hypothesis that sending a connection request with a message works much better (all statistics and p-values support this and make it statistically significant).\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Faorosoeon%2Flinkedin_statistical_inference","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Faorosoeon%2Flinkedin_statistical_inference","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Faorosoeon%2Flinkedin_statistical_inference/lists"}