### Purpose - To build out a basic AI model to enable companies to predict churn in their application and give them actionable insights on what to do. ### Principles - Quality - We want to be able to give quality results that are actually useful. Want to be sure to create value and provide effective insights that actually solve a problem. ### Outcome Visioning In the next month, we want to be able to build a basic AI model that is able to make a good prediction in terms of which users are likely to churn. Our product can further identify what actions or features of users lead to churn and also prevent them. We are able to say with a degree of confidence that the product works and we're able to test it out with a few companies information. Create the predict churn command that allows you to cohort users and then understand the churn rates for them as well. ### Brainstorming - What do I already know about this project? - I have done some homework in terms of how we want to build out the churn analysis. Pecan has a good example on how to make this work and provides a good baseline to work with. - What don’t I know that I need to find out? - I need to learn more about building a regression and classification models and need to research more into building AI and ML products. - One main issue is how to dynamically create AI models without manual effort. - Who can I talk to who might provide insights? - Dev, Harithas and Karthik are good AI/ML folks to talk to. - What can I read or listen to for relevant ideas? - Should consider taking a course - Check out Hugging Face to see if anyone has built similar projects. ### Actions - [x] Investigate Mixpanel and see if it's possible to model using Mixpanel data ✅ 2023-06-07 - [x] Connect OpenOS to Mixpanel ✅ 2023-06-12 - [x] Set up correct OpenOS events for Mixpanel ✅ 2023-06-15 - [x] Try out VertexAI ✅ 2023-06-15 - [x] Show Vivan the way to set up Mixpanel so he can make a video ✅ 2023-06-14 - [x] Frontend for Mixpanel ✅ 2023-06-14 - [x] Pull data from Mixpanel and establish that it's possible ✅ 2023-06-14 - [x] Connect to Preplaced Data and get a data dump on my computer ✅ 2023-07-05 - [x] Create a user Excel with all the properties that they have directly on Mixpanel ✅ 2023-07-05 - [x] Create a CSV with filtered attributes and events as training data for Vertex ✅ 2023-07-05 - [x] Process data with Azure ML and understand results ✅ 2023-07-10 - [x] Clean data for Preplaced Activation Churn ✅ 2023-07-14 - [x] Create one good model using AzureML ✅ 2023-07-14 - [x] Talk to Likhit about Azure ML models and how to clean data ✅ 2023-07-12 - [x] Cleaning up terminal UI ✅ 2023-07-16 - [x] Creating separate page for predictive analysis ✅ 2023-07-16 - [x] Generating list of users csv files for each day ✅ 2023-07-14 - [x] Updating database schema for data modelling ✅ 2023-07-16 - [x] Creating churn by day graph ✅ 2023-07-16 - [x] Creating boxes with predicted and actual churn ✅ 2023-07-16 - [x] Creating feature importance UI where you can click and see graph ✅ 2023-07-16 - [x] Creating cohort list UI ✅ 2023-07-16 - [x] Colour Scheme for login flow ✅ 2023-07-17 - [x] Colour Scheme for home page ✅ 2023-07-17 - [x] Colour Scheme for integrations page ✅ 2023-07-16 - [x] Write script to store results in database on a day to day ✅ 2023-07-16 - [x] Connecting churn by day graph to actual data in database ✅ 2023-07-16 - [x] Connecting feature importance UI to actual data ✅ 2023-07-16 - [x] Connecting cohort list UI to actual data ✅ 2023-07-16 - [x] Applying filters based on drop downs ✅ 2023-07-17 - [x] Reports page hide ✅ 2023-07-17 - [x] New Data Model form page ✅ 2023-07-16 - [x] Text changes across the application ✅ 2023-07-17 - [x] Status bar at the top with respect to data sources ✅ 2023-07-17 - [x] Testing with preplaced data ✅ 2023-07-17 - [x] Connect APIs for New Model page ✅ 2023-07-18 - [x] Scatter plot x and y axis not working properly ✅ 2023-07-18 - [x] Create new preplaced models ✅ 2023-07-18 - [x] New models for Preplaced ✅ 2023-07-23 - [x] Allowing download of users in cohort list table ✅ 2023-07-23 - [x] Change from cube loading spinner in terminal ✅ 2023-07-24 - [x] Fix scatter plot bucketing ✅ 2023-07-24 - [x] Selection filters not working ✅ 2023-07-24 - [x] Add Google Analytics frontend ✅ 2023-07-24 - [x] Loading spinners ✅ 2023-07-24 - [x] Add Selection time period weekly ✅ 2023-07-25 - [x] Download table updates ✅ 2023-07-25 - [x] Create new page for insights ✅ 2023-07-26