def __init__(self, initial_state, total_steps=1000):
"""
:initial_state: the initial graph partition. Must have a cut_edges updater
:total_steps: (defaults to 1000) the total number of steps that the random walk
should perform.
"""
if not initial_state['cut_edges']:
raise ValueError(
'BasicChain needs the Partition to have a cut_edges updater.')
if not initial_state['population']:
raise ValueError(
'BasicChain needs the Partition to have a population updater.')
population_constraint = within_percent_of_ideal_population(
initial_state, 0.01)
compactness_limit = L1_reciprocal_polsby_popper(initial_state)
compactness_constraint = UpperBound(L1_reciprocal_polsby_popper,
compactness_limit)
validator = Validator(default_constraints +
[population_constraint, compactness_constraint])
super().__init__(propose_random_flip,
validator,
always_accept,
initial_state,
total_steps=total_steps)
def write_initial_report(newdir,
outputName,
partition,
df_to_plot,
state_name,
district_col,
num_elections,
election_names,
election_columns,
df,
unique_label,
validator,
county_col=None,
report_entropy=None,
entropy=None,
county_data=None,
reverse_entropy=None):
num_districts = len(partition['cut_edges_by_part'])
with open(outputName, "w") as f:
f.write("<html>\n")
write_header_styles(f)
f.write("<body>\n")
f.write("<h1 width:100%>" + state_name + " Initial Report</h1>\n")
f.write("<h2 width:100%>Initial Data</h2>\n")
f.write("<b>Initial Districting Plan: </b>")
f.write(district_col)
f.write("<br><br><b>Election Data: </b>")
for i in range(num_elections):
f.write(election_names[i] + " ")
df_to_plot.plot(column="initial", cmap='tab20')
plt.axis('off')
plt.title(state_name + " Plan: " + district_col)
plt.savefig(newdir + district_col + "initial.png")
plt.close()
f.write("<div width=100%>\n")
f.write(f" <img src='" + district_col +
"initial.png' width=80%/>\n")
f.write("</div>\n")
f.write("<h2>Statewide Stats</h2>\n\n")
f.write("<b>Number of Districts: </b> " + str(num_districts) + "<br>")
f.write("<b>Population Deviation: </b>" + str(L2_pop_dev(partition)) +
"<br>")
f.write("<b>Reciprocal Polsby-Popper L1: </b>" +
str(L1_reciprocal_polsby_popper(partition)) + "<br>")
f.write("<b>Polsby-Popper L(-1): </b>" +
str(L_minus_1_polsby_popper(partition)) + "<br>")
f.write("<b>Conflicted Edges: </b>" +
str(len(partition["cut_edges"])) + "<br>")
f.write("<b>Boundary Nodes: </b>" +
str(len(partition["boundary_nodes"])) + "<br>")
f.write("<h2> Statewide Partisan Measures</h2>")
f.write("<div width=100%>\n")
for i in range(num_elections):
df["Dperc"] = df[election_columns[i][1]] / (
df[election_columns[i][0]] + df[election_columns[i][1]])
df_to_plot["partisan"] = df_to_plot[unique_label].map(
df.to_dict()["Dperc"])
df_to_plot.plot(column="partisan", cmap="seismic")
plt.axis('off')
plt.title(election_names[i] + " Vote Percentage")
plt.savefig(newdir + "partisan" + str(i) + ".png")
plt.close()
f.write("<img src='partisan" + str(i) + ".png' width=40%/>\n")
f.write("</div>\n")
f.write(
"<table>\n <tr><td>Election</td><td>D Seats</td><td>R Seats</td>" +
"<td> D Votes</td><td>R Votes</td>" +
"<td>D Percent</td><td> R percent</td><td> Mean Median </td>" +
"<td> Efficiency Gap</td> </tr>")
for i in range(num_elections):
f.write("<tr><td>" + election_names[i] + "</td><td>" + str(
how_many_seats_value(partition, election_columns[i][0],
election_columns[i][1])) + "</td><td>" +
str(
how_many_seats_value(partition, election_columns[i][1],
election_columns[i][0])) +
"</td><td>" + str(sum(df[election_columns[i][0]])) +
"</td><td>" + str(sum(df[election_columns[i][1]])) +
"</td><td>" + str(
sum(df[election_columns[i][0]]) /
(sum(df[election_columns[i][0]]) +
sum(df[election_columns[i][1]]))) + "</td><td>" + str(
sum(df[election_columns[i][1]]) /
(sum(df[election_columns[i][0]]) +
sum(df[election_columns[i][1]]))) + "</td><td>" +
str(mean_median(partition, election_columns[i][0] + "%")) +
"</td><td>" + str(
efficiency_gap(partition, election_columns[i][0],
election_columns[i][1])) + "</td></tr>")
f.write("</table>\n\n")
f.write("<h2> District Partisan Measures</h2>")
win_dict = {0: "D", 1: "R"}
dw_list = []
f.write("<div width=100%>\n")
for i in range(num_elections):
district_winners = {}
for j in range(num_districts):
district_winners[j + 1] = int(
partition[election_columns[i][1]][j + 1] > partition[
election_columns[i][0]][j + 1])
df_to_plot["district_partisan"] = df_to_plot[district_col].map(
district_winners)
dw_list.append(district_winners)
df_to_plot.plot(column="district_partisan", cmap="seismic")
plt.axis('off')
plt.title(election_names[i] + " District Winners")
plt.savefig(newdir + district_col + "district_partisan" + str(i) +
".png")
plt.close()
f.write("<img src='" + district_col + "district_partisan" +
str(i) + ".png' width=40%/>\n")
f.write("</div>\n")
f.write("<table>\n <tr><td>District</td>")
for i in range(num_elections):
f.write("<td>" + election_names[i] + " Winner</td>" + "<td>" +
election_names[i] + " Win %</td>")
f.write("</tr>")
for i in range(num_districts):
f.write("<tr><td>" + str(i + 1) + "</td>")
for j in range(num_elections):
f.write("<td>" + win_dict[dw_list[j][i + 1]] + "</td><td>" +
str(partition[election_columns[j][dw_list[j][i + 1]] +
"%"][i + 1]) + "</td>")
f.write("</tr></table>")
f.write("</table>")
f.write("<h2> District Measures </h2>")
df_to_plot["PP"] = df_to_plot[district_col].map(
partition["polsby_popper"])
df_to_plot.plot(column="PP", cmap="cool")
plt.axis("off")
plt.title("District Polsby-Popper Scores")
plt.savefig(newdir + district_col + "district_PP" + ".png")
plt.close()
f.write("<div width=100%>\n")
f.write(f" <img src='" + district_col +
"district_PP.png' width=80%/>\n")
f.write("</div>\n")
f.write(
"<table>\n<td>District</td><td>Units</td><td>Conflicted Edges</td>"
+ "<td>Population Deviation %</td><td>Polsby-Popper</td></tr>")
total_population = sum(partition['population'].values())
mean_population = total_population / num_districts
for i in range(num_districts):
f.write("<tr><td>" + str(i + 1) + "</td><td>" +
str(len(partition.parts[i + 1])) + "</td><td>" +
str(len(partition["cut_edges_by_part"][i + 1])) +
"</td><td>" +
str((partition["population"][i + 1] - mean_population) /
(mean_population)) + "</td><td>" +
str(partition["polsby_popper"][i + 1]) + "</td></tr>")
node_lengths = [len(x) for x in partition.parts.values()]
f.write(
"<tr><td> Mean</td><td>" + str(sum(node_lengths) / num_districts) +
"</td><td>" + str(
sum([len(x)
for x in partition["cut_edges_by_part"].values()]) /
num_districts) + "</td><td>" + str(
sum([(x - mean_population) / mean_population
for x in partition["population"].values()]) /
num_districts) + "</td><td>" + str(
sum([x for x in partition["polsby_popper"].values()]) /
num_districts) + "</td></tr>")
f.write(
"<tr><td> Max</td><td>" +
str(max([len(x)
for x in partition.parts.values()])) + "</td><td>" +
str(max([len(x)
for x in partition["cut_edges_by_part"].values()])) +
"</td><td>" + str(
max([(x - mean_population) / mean_population
for x in partition["population"].values()])) +
"</td><td>" +
str(max([x for x in partition["polsby_popper"].values()])) +
"</td></tr>")
f.write(
"<tr><td> Min</td><td>" +
str(min([len(x)
for x in partition.parts.values()])) + "</td><td>" +
str(min([len(x)
for x in partition["cut_edges_by_part"].values()])) +
"</td><td>" + str(
min([(x - mean_population) / mean_population
for x in partition["population"].values()])) +
"</td><td>" +
str(min([x for x in partition["polsby_popper"].values()])) +
"</td></tr>")
f.write("</table>")
if report_entropy is not None:
f.write("<h2>Entropy Report</h2>")
df_to_plot["county"] = df_to_plot[unique_label].map(
df.to_dict()[county_col])
df_to_plot.plot(column="county", cmap="tab20")
plt.axis("off")
plt.title(state_name + " County Map")
plt.savefig(newdir + "counties" + ".png")
plt.close()
f.write("<div width=100%>\n")
f.write(f" <img src='" + district_col +
"initial.png' width=40%/>\n")
f.write(f" <img src='counties.png' width=40%/>\n")
f.write("</div>\n")
f.write("<b>Number of Counties: </b>")
f.write(str(len(county_data)))
f.write("<br><b>Number of Split Counties: </b>")
f.write(str(sum([x[1] for x in county_data])))
f.write("<br><b>4/5 function regular weight: </b>")
f.write(str(entropy[0][0]))
f.write("<br><b>4/5 function inverse weight: </b>")
f.write(str(entropy[0][1]))
f.write("\n")
f.write("<br><b>4/5 function no weight: </b>")
f.write(str(entropy[0][2]))
f.write("\n")
f.write("<br><b>Linear function regular weight: </b>")
f.write(str(entropy[1][0]))
f.write("\n")
f.write("<br><b>Linear function inverse weight: </b>")
f.write(str(entropy[1][1]))
f.write("\n")
f.write("<br><b>Linear function no weight: </b>")
f.write(str(entropy[1][2]))
f.write("\n")
f.write("<br><b>Shannon function regular weight: </b>")
f.write(str(entropy[2][0]))
f.write("\n")
f.write("<br><b>Shannon function inverse weight: </b>")
f.write(str(entropy[2][1]))
f.write("\n")
f.write("<br><b>Shannon function no weight: </b>")
f.write(str(entropy[2][2]))
f.write("<h2>Splits by County</h2>")
f.write(
"<table><tr><td>County Label</td><td> Split?</td><td>Population</td></tr>"
)
yn_dict = {0: "No", 1: "Yes"}
for i in county_data:
f.write("<tr><td><b>County" + str(i[0]) + "</b></td><td>" +
yn_dict[i[1]] + "</td><td>" + str(i[2]) + "</td></tr>")
if len(i[3]) > 1:
for j in range(len(i[3])):
f.write("<tr><td> Part " + str(j) +
"</td><td></td><td>" + str(i[3][j]) +
"</td></tr>")
f.write("</table>")
f.write("<h2>District Splits by County</h2>")
f.write("<b>4/5 function regular weight: </b>")
f.write(str(reverse_entropy[0][0]))
f.write("<br><b>4/5 function inverse weight: </b>")
f.write(str(reverse_entropy[0][1]))
f.write("\n")
f.write("<br><b>4/5 function no weight: </b>")
f.write(str(reverse_entropy[0][2]))
f.write("\n")
f.write("<br><b>Linear function regular weight: </b>")
f.write(str(reverse_entropy[1][0]))
f.write("\n")
f.write("<br><b>Linear function inverse weight: </b>")
f.write(str(reverse_entropy[1][1]))
f.write("\n")
f.write("<br><b>Linear function no weight: </b>")
f.write(str(reverse_entropy[1][2]))
f.write("\n")
f.write("<br><b>Shannon function regular weight: </b>")
f.write(str(reverse_entropy[2][0]))
f.write("\n")
f.write("<br><b>Shannon function inverse weight: </b>")
f.write(str(reverse_entropy[2][1]))
f.write("\n")
f.write("<br><b>Shannon function no weight: </b>")
f.write(str(reverse_entropy[2][2]))
for i in range(num_elections):
updaters = {
**updaters,
**votes_updaters(election_columns[i], election_names[i])
}
# This builds the partition object
initial_partition = Partition(graph, assignment, updaters)
# Desired validators go here
# Can change constants and bounds
pop_limit = .01
population_constraint = within_percent_of_ideal_population(
initial_partition, pop_limit)
compactness_limit_L1 = 1.01 * L1_reciprocal_polsby_popper(initial_partition)
compactness_constraint_L1 = UpperBound(L1_reciprocal_polsby_popper,
compactness_limit_L1)
compactness_limit_Lm1 = .99 * L_minus_1_polsby_popper(initial_partition)
compactness_constraint_Lm1 = LowerBound(L_minus_1_polsby_popper,
compactness_limit_Lm1)
validator = Validator([
refuse_new_splits, no_vanishing_districts, single_flip_contiguous,
population_constraint, compactness_constraint_Lm1
])
# Names of validators for output
# Necessary since bounds don't have __name__'s
list_of_validators = [
Tally(area_col, alias='areas'),
'polsby_popper':
polsby_popper,
'cut_edges_by_part':
cut_edges_by_part
}
# This builds the partition object
initial_partition = Partition(graph, assignment, updaters)
# Desired validators go here
pop_limit = .02
population_constraint = within_percent_of_ideal_population(
initial_partition, pop_limit)
compactness_limit = L1_reciprocal_polsby_popper(initial_partition)
compactness_constraint = UpperBound(L1_reciprocal_polsby_popper,
compactness_limit)
validator = Validator([
refuse_new_splits, no_vanishing_districts, single_flip_contiguous,
population_constraint, compactness_constraint
])
# Add cyclic updaters :(
# updaters['metagraph_degree'] = MetagraphDegree(validator, "metagraph_degree")
# This builds the partition object (again) :(
# initial_partition = Partition(graph, assignment, updaters)
print("setup chain")