Machine learning task / google colab

you will work on redistricting the Congressional map for New Jersey.

You are also provided with the map for New Hampshire as it is a smaller state, with only two districts, that you can use as a sandbox. Various functions and algorithms for NH will be provided to you in recitation ([login to view URL]). You can base your work on NJ on these. We also provide a NJ-centric Python notebook, with an example map, as a starting point. ([login to view URL]).

Your goal is to produce two maps:

One map should be as fair as possible

One map should be designed to give a party/community/area some advantage (unfair map) using some creative gerrymandering techniques. The less likely the map would be to be thrown out of courts for obvious gerrymandering, the better.

The NJ State Assembly has 12 districts. You can see the current (2022) map here: [login to view URL]::6ff0b024-2e5d-4e9b-ba0e-56f6de17ea80Links to an external site.

We are providing you with various datasets and tools to assist you. All are in Canvas>Files>Assignments>Redistricting

A pdf that provides a tutorial to the DRA web app (DRA [login to view URL])

A file with demographics and electoral data for each NJ and NH precincts (Map Data/[login to view URL] and Map Data/[login to view URL]). Each precinct is identified by its GEOID20Links to an external site. key. You can see a description of the fields in [login to view URL]

Directories with the shape of all precincts, using TIGER shape data (nh_vtd_2020_bound and nj_vtd_2020_bound). The provided notebooks give examples of how to use the shape data. You can also use the pre-computed precinct Contiguity indexes: [login to view URL] and [login to view URL]

Deliverables

Code (in Python Notebook format) used to generate your districts

csv files with the two assignments

In-depth analyis of your assignments, your choices, and what that means in terms of the various metrics we discussed in class.

1. Code

While DRA allows you to "paint" maps, this is a CS algorithm course and you are required to have some algorithmic component (and code) used to generate the maps. The maps have to satisfy two constraints: contiguous districts and balanced population (+-10%). In terms of algorithms, you have several options, as we saw in class:

Build a new map from scratch, through splitting, randomization, optimization, or enumeration (the latter is probably not a good idea).

Start from a seed map (for instance you can "paint" no more than 20% of the districts to say keep some communities of interest together) and build your map from there algorithmically (e.g. flood fill).

Start from the current map and modify it. Your result map should be significantly different (10%).

2. CSV file

This is pretty self-explanatory. Your files should assign precincts to districts. There should be 6361 precincts (rows). You can see examples of a correct csv in Recitation Maps

Your CSV file should be in a format that allows you to import it into DRA.

3. Data Analysis

You are not required to write data analysis code: DRA does it for you. Upload your maps in DRA and you will be able to see various metrics.

You should submit a written analysis of these metrics and their implications for your map. How does your map(s) compare to the current assignment? What are your map(s)'s strengths and weaknesses?

and DRA tutorial on google colab link. Send the link to them to take a look on how

so basically i need 2 things like how to run the maps and save the maps as a csv on DRA

I thought the guy have done this part already since this is kinda part of it

it shows how to create a map on DRA

Python Machine Learning (ML) Software Architecture Java C Programming

Project ID: #37460157

About the project