"""
Library for visualizing model output
These functions are primarily graphing functions, with some supporting functions as well
"""
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from scipy import stats
import seaborn as sns
from utils import N_ROUNDS
def groupby_f_data(f_data, colname, bins):
"""
group by filtered data with player outcome and calculate the win percentage
colname will be either 'player_outcome' or 'agent_outcome' for plotting human or agent results
"""
modified_f_data = f_data.dropna()
labs = [str(int(round(a * (N_ROUNDS / bins), 0))) for a in range(1, bins + 1)]
modified_f_data['bin'] = pd.cut(modified_f_data.loc[:, ('round_index')], bins, labels = labs)
grouped_data = modified_f_data[['bot_strategy', 'player_id','bin', colname]].groupby(
['bot_strategy', 'player_id', 'bin'])[colname].value_counts('count').rename('pct').reset_index()
return grouped_data
def win_summary(grouped_data, colname):
"""
filter out the win data and add mean, SD, and SEM
colname will be either 'player_outcome' or 'agent_outcome' for plotting human or agent results
"""
win_data = grouped_data[grouped_data[colname] == 'win'].reset_index()
win_summary = win_data[['bot_strategy', 'bin', 'pct']].groupby(
['bot_strategy', 'bin'])['pct'].agg(
[np.mean, np.std, stats.sem]).reset_index()
return win_summary
def plot_win_rates(data, img_name=None):
"""
generate plot displaying win rates against each bot, binned by rounds
"""
data['bot_strategy'] = data['bot_strategy'].replace([
'prev_move_positive', 'prev_move_negative',
'opponent_prev_move_positive', 'opponent_prev_move_nil',
'win_nil_lose_positive', 'win_positive_lose_negative',
'outcome_transition_dual_dependency'
],
[
'Previous move (+)', 'Previous move (-)',
'Opponent previous move (+)', 'Opponent previous move (0)',
'Win-stay-lose-positive', 'Win-positive-lose-negative',
'Outcome-transition dual dependency'
])
data['bin'] = data['bin'].replace(['1', '2', '3', '4', '5', '6', '7', '8', '9', '10'], ['100', '200', '300', '400', '500', '600', '700', '800', '900', '1000'])
palette = {
'Opponent previous move (0)':'#A569BD',
'Opponent previous move (+)':'#8E44AD',
'Outcome-transition dual dependency':'#A04000',
'Previous move (-)':'#85C1E9',
'Previous move (+)':'#2874A6',
'Win-stay-lose-positive':'#A9DFBF',
'Win-positive-lose-negative':'#229954'
}
hue_order = ['Previous move (-)','Previous move (+)','Opponent previous move (0)','Opponent previous move (+)',
'Win-stay-lose-positive','Win-positive-lose-negative','Outcome-transition dual dependency']
f, ax = plt.subplots(figsize=(15, 10))
g = sns.pointplot(
x = "bin", y = "pct", hue = "bot_strategy", scale = 2.5,
palette=palette, s = 400, ax = ax, data = data,hue_order = hue_order)
plt.ylim(0, 1.0)
ax.set_xticklabels(["30","","","","150","","","","","300"])
ax.set_yticks([0,0.1,0.2,0.3,0.4, 0.5,0.6,0.7,0.8,0.9,1.0])
ax.set_yticklabels(["0","","","","","0.5","","","","","1"])
plt.xlabel('Trial round',fontdict={'fontsize':25},fontweight="bold")
plt.ylabel('Mean win percentage',fontdict={'fontsize':25},fontweight="bold")
plt.axhline(y = 1/3, color = 'grey', linestyle = '--',linewidth = 5)
for label in (ax.get_xticklabels()+ax.get_yticklabels()):
label.set_fontsize(15)
label.set_fontweight("bold")
handles, labels = ax.get_legend_handles_labels()
ax.legend(handles=handles[1:], labels=labels[1:])
ax.legend(fontsize = 15)
sns.despine()
if img_name:
plt.savefig(os.path.join('img', img_name), dpi=300, bbox_inches='tight', transparent=True)
return g
