workout-app

Simple workout scheduler built for the course "Introduction to Data Science" in University of Helsinki.

Problem description

The user wants to schedule $n$ workouts a given timeframe of available timeslots, i.e., starting date and hour of a workout. Given historical observations of the "crowdedness" under some metric (e.g., number of people arrived in a gym), how should the user choose the workout times in order to avoid the busy time slots? Other factors can also be taken in account in the preferences of the user, like the distance to the workout location and so on.

Approach to solution

Given predictions as time series, form predictions for future values of "crowdnedness". This leads to a combinatorial problem, where the user wants to find to the optimal time slots to minimize total "crowdedness" while meeting the constraint of total required number of workouts. We model this binary optimization problem as a weighted MAX-SAT problem, which can be solved using existing Python libraries, such as PySAT.

Scheduler workflow

flowchart TD
    A[User] -->|Clicks on Scheduler Calendar| B[Client]
    B -->|Sends chosen schedule| C[Backend]
    C -->|Receives schedule| D[Predict Class]
    D -->|Fetches time series prediction| E[Solver Class]
    E -->|Finds MAXSAT solution| F[Solution]
    F -->|Sends solution back| B
    B -->|Displays solution to user| A

Loading

Setting up the environment

Install dependencies and start backend in ./backend

poetry install
poetry run python app.py &

Install dependencies and start frontend in ./frontend

npm install && npm run dev &