In [1]:
import scipy.io as scio
# Import .mat file as python object
data = scio.loadmat('../data/data_18333.mat', struct_as_record=False)
For now, let's focus on the following:
This is similar to examples/hddm_simple.csv, used to play around with the HDDM library
In [2]:
dat_struct = data['data'][0,0] # Actual data structure, owns to matlab weirdness
Before outputing to CSV, data for each subject will go in a python dictionary in the form of key --> array. The plan is to then create an array of dictionaries for all patients, with each dictionary representing the data gathered for an individual
In [3]:
"""
Conversion from convoluted numpy array that scipy.io spits out to a more
pythonic data structure.
Leverage python instead of numpy for data manipulation, since the use of
numpy isn't really necessary for this data.
"""
#subject['rt'] = dat_struct.rt1.tolist()[0]
#subject['response'] = [x[0] for x in dat_struct.perf1.tolist()]
#subject['stim'] = [x[0] for x in dat_struct.conditions1.tolist()]
csv_keys = ['rt', 'response', 'stim']
reaction_times = dat_struct.rt1.tolist()[0]
responses = [x[0] for x in dat_struct.perf1.tolist()]
stimuli = [x[0] for x in dat_struct.conditions1.tolist()]
subject = []
for exp_run in list(zip(reaction_times, responses, stimuli)):
trial = dict.fromkeys(csv_keys)
trial['rt'], trial['response'], trial['stim'] = exp_run
subject.append(trial)
Now that the data is in a desirable format, we can dump it to a CSV file
In [4]:
import csv
with open('../data/data_18333.csv', 'w') as f:
w = csv.DictWriter(f, csv_keys)
w.writeheader()
w.writerows(subject)
In [5]:
keys = ['rt', 'response', 'stim']
In [6]:
def mat2py(mat_path):
"""
Function to convert mat file to a pythonic data structure
Returns list of dictionaries mapping to spectific attributes
"""
data = scio.loadmat(mat_path, struct_as_record=False)
dat_struct = data['data'][0,0]
reaction_times = dat_struct.rt1.tolist()[0]
responses = [x[0] for x in dat_struct.perf1.tolist()]
stimuli = [x[0] for x in dat_struct.conditions1.tolist()]
subject = []
for exp_run in list(zip(reaction_times, responses, stimuli)):
trial = dict.fromkeys(keys)
trial['rt'], trial['response'], trial['stim'] = exp_run
subject.append(trial)
return subject
In [7]:
def subject2csv(subject, mat_path):
csv_path = mat_path.replace('.mat', '.csv')
print(csv_path)
with open(csv_path, 'w') as f:
w = csv.DictWriter(f, keys)
w.writeheader()
w.writerows(subject)
In [8]:
"""
Iterate through all .mat files in the data directory and
convert them to csv format
"""
import glob
data_dir = '../data/pilot_subjects/'
mat_files = glob.glob(str(data_dir) + '*.mat')
for mat in mat_files:
subject2csv(mat2py(mat), mat)
A cleaned-up version of the above is found under utils/matparser.py
In [10]:
dat_struct.__dict__
Out[10]: