Skip to content

This is the solution for a challenge proposed by Citric Sheep, aiming to create an intelligent system for elevators.

Notifications You must be signed in to change notification settings

teixeirazeus/Smart-Elevator-Project

 
 

Repository files navigation

banner

Smart Elevator Project

This is the solution for a challenge proposed by Citric Sheep, aiming to create an intelligent system for elevators. You can have a more detailed description of the challenge at the end of this readme.

The solution

The intelligent system aims to choose the optimal floor for the elevator to rest, based on the elevator's call history. The system consists of a server that receives elevator calls and a database that stores these calls.

For this purpose, a heuristic algorithm has been proposed. Based on the call history, the algorithm selects the floor that has been called the most within a one-hour interval. Minimizing the following cost function:

$$\text{cost(resting\_floor)} = \sum_{i=1}^{n} \text{calls}_{x_i} \times \lvert x_i - \text{resting\_floor} \rvert$$

Where $\text{calls}_{x_i}$ is the number of calls for floor $x_i$ within the one-hour interval, and $n$ is the total number of floors.

The algorithm tests all possible floors and chooses the one that minimizes the cost function

For example, consider the following call history:

1, 1, 1, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 5

Organizing it into a call vector:

[3, 5, 6, 5, 1]

The algorithm tests all possible floors and chooses the one that minimizes the cost function. In this case, floor 3 (index 2) is the one that minimizes the cost function.

The minimization strategy over a one-hour timeframe enables the algorithm to anticipate trends throughout the day. For example, in a residential building elevator, there might be an increased demand around 6 p.m. as people return from work. Similarly, in a commercial building, there could be a shift of people from floor A to floor B during lunchtime.

How to run

Setting up the environment

Create a virtual environment

python3.11 -m venv env

Activate the virtual environment

source env/bin/activate

Install the requirements

pip install -r requirements.txt

Running the tests

pytest .

Running the application

python main.py

Running the server

python server.py

You can test the server endpoints using a collection of requests in the request_collection.json file. You can use programs like Insomnia or Postman.

insomia

The Challenge Description

When an elevator is empty and not moving this is known as it's resting floor. The ideal resting floor to be positioned on depends on the likely next floor that the elevator will be called from.

We can build a prediction engine to predict the likely next floor based on historical demand, if we have the data.

The goal of this project is to model an elevator and save the data that could later be used to build a prediction engine for which floor is the best resting floor at any time

  • When people call an elevator this is considered a demand
  • When the elevator is vacant and not moving between floors, the current floor is considered its resting floor
  • When the elevator is vacant, it can stay at the current position or move to a different floor
  • The prediction model will determine what is the best floor to rest on

The requirement isn't to complete this system but to start building a system that would feed into the training and prediction of an ML system

You will need to talk through your approach, how you modelled the data and why you thought that data was important, provide endpoints to collect the data and a means to store the data. Testing is important and will be used verify your system

In short

This is a domain modeling problem to build a fit for purpose data storage with a focus on ai data ingestion

  • Model the problem into a storage schema (SQL DB schema or whatever you prefer)
  • CRUD some data
  • Add some flair with a business rule or two
  • Have the data in a suitable format to feed to a prediction training algorithm

To start

  • Fork this repo and begin from there
  • For your submission, PR into the main repo. We will review it, a offer any feedback and give you a pass / fail if it passes PR
  • Don't spend more than 4 hours on this. Projects that pass PR are paid at the standard hourly rate

Marking

  • You will be marked on how well your tests cover the code and how useful they would be in a prod system
  • You will need to provide storage of some sort. This could be as simple as a sqlite or as complicated as a docker container with a migrations file
  • Solutions will be marked against the position you are applying for, a Snr Dev will be expected to have a nearly complete solution and to have thought out the domain and built a schema to fit any issues that could arise A Jr. dev will be expected to provide a basic design and understand how ML systems like to ingest data

Trip-ups from the past

Below is a list of some things from previous submissions that haven't worked out

  • Built a prediction engine
  • Built a full website with bells and whistles
  • Spent more than the time allowed (you won't get bonus points for creating an intricate solution, we want a fit for purpose solution)
  • Overcomplicated the system mentally and failed to start

About

This is the solution for a challenge proposed by Citric Sheep, aiming to create an intelligent system for elevators.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 98.1%
  • Shell 1.9%