Skip to content

Latest commit

 

History

History
84 lines (73 loc) · 5.16 KB

A2Deliverable.md

File metadata and controls

84 lines (73 loc) · 5.16 KB
Raw

A2 Deliverable

October 21, 2022

This document contains all the core expectations for assignment 2.

Description of the features our project intends to implement

CI

  • We prioritized this because "everything that can be automated at this point must be".
  • Deliverables:
    • Create Github action(s) to automatically run our testing suite on commit to any branch.
      • Assigned to Kirill.
      • Due on November 9, 2022.
      • Acceptance criteria:
        • Commiting to a branch runs the testing suite.
    • Create a Github action to delete feature branches after they've been merged to dev.
      • Assigned to Kirill.
      • Due on November 9, 2022.
      • Acceptance criteria:
        • After merging a branch to dev, that branch must no longer exist.

We will also be implementing our first feature: a song recommendation engine.
As a user, I want to select a list of songs I like and have new songs recommended to me.

ML Model

  • We prioritized this because it is the backbone of our recommendation feature. Without the ML model, we can't generate recommendations.
  • Deliverables:
    • Set up the ML model and get it running properly.
      • Assigned to Roger.
      • Due on October 26, 2022.
      • Acceptance criteria:
        • Given a list of input track_uris, running the ML model generates a CSV of recommended track_uris that make sense (based on sanity check).
        • Code for the ML model must be commited to the repo without errors.
    • Create a black-box public function (interface) that takes in a list of 1-5 track_uris and returns a list of 100 recommended track_uris.
      • Assigned to Roger.
      • Due on November 2, 2022.
      • Acceptance criteria:
        • Given a list of input track_uris, the black-box function successfully uses the ML model to generate a new list of recommended songs.
        • The function is able to be called by the backend endpoint.

Backend Endpoint

  • We prioritized this because the backend needs to turn the ML model's recommended track_uris into a readable format. If the frontend just received a list of random characters (aka track_uris), it wouldn't be very useful to the user.
  • Deliverables:
    • Create a POST endpoint that takes a request with a list of input track_uris and returns a response with a list of track data. The request input will be a JSON of { "data": list<string> } and the response will be a JSON of list<{ "name": string, "artists": list<string>, "image_url": string }>.
      • Assigned to Eddie.
      • Due on October 26, 2022.
      • Acceptance criteria:
        • Endpoint returns a response of the correct format.
        • Endpoint calls the ML model black-box interface function to get a list of recommended track_uris.
        • Endpoint uses the Spotify track API to query name, artists, and cover image of these recommended track_uris to send to the frontend.

Frontend Component

  • We prioritized this because we need a way to present our recommendations to the user in a recognizable way. As mentioned before, handing the users a list of random characters (aka. track_uris) would not be very useful to them.
  • Deliverables:
    • An input box (for the user to type song names) and a search button (to generate recommended songs).
      • Assigned to Burt and Kirill.
      • Due on October 26, 2022.
      • Acceptance criteria:
        • Input box and search button need to be usable by the user.
        • Input box needs to recommend autofill options for song names using Spotify searching API.
        • Search button needs to call the corresponding backend endpoint and pass it the list of songs from the input box.
    • Display recommended songs for the user.
      • Assigned to Kirill.
      • Due on October 26, 2022.
      • Acceptance criteria:
        • Need to write UI components to conditionally render responses from the recommendation backend endpoint.
        • These UI components need to be somewhat readably by the user.\

Progress towards our Milestones

Milestone 2: Enforce automation and testing.

  • Ensure our code is linted, styled, and sufficiently tested. IN PROGRESS
  • Set up automatic testing, building, integration and deployment pipelines. DONE

Milestone 3: Create a backend that runs machine learning on our data to answer interesting, complex questions.

  • Set up ML models that work on our data. IN PROGRESS
  • Code logic to process the info our ML models output and communicate it to the frontend. IN PROGRESS (first feature done)

Milestone 4: Create a client that allows users to input queries they have related to our data and display meaningful conclusions and visualizations about it.

  • Design the user interface, with a focus on exposing all features, handling query input, and displaying data in appropriate formats. DONE
  • Implement UI components. IN PROGRESS (first feature done)
  • Implement communication with backend. IN PROGRESS (first feature done)

Blameless Postmortem on Assignment 1

We have created a Blameless Postmortem to reflect on our experience working on A1 and how we can improve our processes for future deliverables.