def plot_pulled_prob_MF_agent(data_path):
subjectd_IDs = get_directory_names(data_path)
for id in subjectd_IDs:
repeat, common_rewarded, unique_rewarded = extract_MF_behavior_data(data_path, id)
common1_unique0, common0_unique1, common0_unique0, common1_unique1 = \
find_repeat_probs_MF(repeat, common_rewarded, unique_rewarded)
fig = plt.figure('MF agent repeat probabilities')
xs = [0, 1]
ys = [common1_unique0, common1_unique1]
plt.scatter(xs, ys, c='b', linewidths=0.5, marker='.', s=3)
ys = [common0_unique0, common0_unique1]
plt.scatter(xs, ys, c='r', linewidths=0.5, marker='.', s=3)
plt.xticks(range(2), ['U=0', 'U=1'])
plt.ylim([0, 1])
mean_c1u1, std_c1u1, mean_c1u0, std_c1u0, mean_c0u1, std_c0u1, mean_c0u0, std_c0u0 = \
extract_MF_probabilities(data_path)
plt.errorbar(xs, [mean_c0u0, mean_c0u1], yerr=[std_c0u0, std_c0u1], label='common 0')
plt.errorbar(xs, [mean_c1u0, mean_c1u1], yerr=[std_c1u0, std_c1u1], label='common 1')
plt.legend()
plt.ylabel('Prob(repeat)')
plt.show()
def extract_data_MF_glmfit(data_path):
pulled_dataframe = pd.DataFrame({
'repeat': [],
'common_reward': [],
'unique_reward': [],
'subject_id': []
})
subject_models, subject_results, pulled_model, pulled_results = [], [], None, None
subjectd_IDs = get_directory_names(data_path)
for id in subjectd_IDs:
model, result, subject_df = glmfit_for_single_subject(data_path, id)
subject_models.append(model)
subject_results.append(result)
pulled_dataframe = pulled_dataframe.append(subject_df,
ignore_index=True)
pulled_model, pulled_results = mixedlm(
pulled_dataframe, 'repeat ~ common_reward*unique_reward', {
'a': '0+C(subject_id)',
'b': '0+C(subject_id)*common_reward',
'c': '0+C(subject_id)*unique_reward'
})
return subject_models, subject_results, pulled_model, pulled_results
def _find_new_subject_id_():
dir_names = get_directory_names(TaskParams.data_dir)
if dir_names:
subject_ids = [int(name) for name in dir_names]
return max(subject_ids) + 1
else:
return 1
def extract_MB_probabilities(data_path):
subjectd_IDs = get_directory_names(data_path)
c1s, c0s = [], []
for id in subjectd_IDs:
repeat, common_rewarded, _ = extract_MB_behavior_data(data_path, id)
common1, common0 = find_repeat_probs(repeat, common_rewarded)
c1s.append(common1)
c0s.append(common0)
return np.mean(c1s), sem(c1s), np.mean(c0s), sem(c0s)
def extract_MF_probabilities(data_path):
subjectd_IDs = get_directory_names(data_path)
c1u1s, c1u0s, c0u1s, c0u0s = [], [], [], []
for id in subjectd_IDs:
repeat, common_rewarded, unique_rewarded = extract_MF_behavior_data(
data_path, id)
common1_unique0, common0_unique1, common0_unique0, common1_unique1 = \
find_repeat_probs(repeat, common_rewarded, unique_rewarded)
c1u1s.append(common1_unique1)
c1u0s.append(common1_unique0)
c0u1s.append(common0_unique1)
c0u0s.append(common0_unique0)
return np.mean(c1u1s), sem(c1u1s), np.mean(c1u0s), sem(c1u0s),\
np.mean(c0u1s), sem(c0u1s), np.mean(c0u0s), sem(c0u0s)
def plot_pulled_prob_MB_agent(data_path):
subjectd_IDs = get_directory_names(data_path)
for id in subjectd_IDs:
repeat, common_rewarded, _ = extract_MB_behavior_data(data_path, id)
common1, common0 = find_repeat_probs_MB(repeat, common_rewarded)
fig = plt.figure('MB agent repeat probabilities')
xs = [0, 1]
ys = [common0, common1]
plt.scatter(xs, ys, c='b', linewidths=0.5, marker='.', s=3)
plt.xticks(range(2), ['C=0', 'C=1'])
plt.ylim([0, 1])
mean_c1, std_c1, mean_c0, std_c0 = extract_MB_probabilities(data_path)
plt.errorbar(xs, [mean_c0, mean_c1], yerr=[std_c0, std_c1])
plt.ylabel('Prob(repeat)')
plt.show()