US vote trends

Load packages.

import pandas as pd import plotly.express as px

Data have been scraped from the New York Times by Alex Gaynor, and continuously updated on his GitHub repository.

elec = pd.read_csv("https://raw.githubusercontent.com/alex/nyt-2020-election-scraper/master/battleground-state-changes.csv")

Convert the time column.

elec.timestamp = pd.to_datetime(elec.timestamp, format="%Y-%m-%d %H:%M:%S.%f").dt.round(freq = "s")

Data are shown as leading candidate, trailing candidate and votes for the leader and the trailer. I had to apply some filters to get the votes for Biden and Trump.

elec["votes_biden"] = 0 elec["votes_trump"] = 0 filter_trump_leads = elec["leading_candidate_name"] == "Trump" filter_biden_leads = elec["leading_candidate_name"] == "Biden" # where trump leads elec.loc[filter_trump_leads, "votes_trump"] = elec.loc[filter_trump_leads, "leading_candidate_votes"] elec.loc[filter_trump_leads, "votes_biden"] = elec.loc[filter_trump_leads, "trailing_candidate_votes"] # where biden leads elec.loc[filter_biden_leads, "votes_trump"] = elec.loc[filter_biden_leads, "trailing_candidate_votes"] elec.loc[filter_biden_leads, "votes_biden"] = elec.loc[filter_biden_leads, "leading_candidate_votes"] # remove zeroes elec = elec.loc[(elec.votes_biden != 0) | (elec.votes_trump != 0), :]

Plot all the remaining states.

px.line( elec, x='timestamp', y=["votes_biden", "votes_trump"], facet_col="state", facet_col_wrap=1, width=600, height=600 ).update_yaxes( matches=None )

Plot one state at the time.

states = elec.state.unique() states

state_rank = 2 px.line( elec.loc[elec.state == states[state_rank], :], x='timestamp', y=["votes_biden", "votes_trump"], title = states[state_rank] )

US vote trends

Recommended on Deepnote

Stock market analysis

10 ways to create NumPy arrays

Wide Residual Networks