Looping through nested dictionary values to populate second dictionary - python

Question

I have the following dictionary:

rts = {
         "PO1": {
                   "congruent": {

                             "rt": [0.647259, 0.720116, 0.562909, 0.538918, 0.633367],
                        "correct": ["True", "True", "True", "True", "True", "False",]
                             },

                  "incongruent": {

                             "rt": [0.647259, 0.720116, 0.562909, 0.538918, 0.633367],
                        "correct": ["True", "True", "True", "True", "True", "False",]

                             }
         },
         "PO2": {
                   "congruent": {

                             "rt": [0.647259, 0.720116, 0.562909, 0.538918, 0.633367],
                        "correct": ["True", "True", "True", "True", "True", "False",]
                             },

                  "incongruent": {

                             "rt": [0.647259, 0.720116, 0.562909, 0.538918, 0.633367],
                        "correct": ["True", "True", "True", "True", "True", "False",]

                             }
         }
}

Here is the code I have so far:

import csv
from pathlib import Path
import json
import numpy as np
from numpy import array


def main():
    rts = {}
    statsDict = {}

    data = Path('C:/Users/oli.warriner/Desktop/data(2)/data')

    for csvfile in data.glob('*.csv'):
        key = csvfile.stem

        with csvfile.open() as f:
            csv_reader = csv.reader(f)

            # Skip the header
            _ = next(csv_reader)

            rts[key] = {
                'congruent': {
                    'rt': [],
                    'correct': []
                },
                'incongruent': {
                    'rt': [],
                    'correct': []
                },
            }

            for tn, ctext, cname, condition, response, rt, correct in csv_reader:
                rts[key][condition]['rt'].append(float(rt))
                rts[key][condition]['correct'].append(correct)


    for k in rts:
        key = k
        statsDict[key] = {
                'congruent': {
                    'mean': [],
                    'stddev': [],
                    'correct': []
                },
                'incongruent': {
                    'mean': [],
                    'stddev': [],
                    'correct': []
                },
            }
        for n in rts[k]:
            for i in rts[key][n]

    array([rts[k] for k in rts]).mean()
    print(array)

if __name__ == "__main__":
    main()

I am reading a directory of csv files to produce the "rts" dictionary you see above (Its much bigger than that I have just shortened for here).

What I am now looking to do is to use the "rts" dictionary to populate the "statsDict".

I need to loop through the "rts" dictionary and calculate the mean and standard deviation from the "rt" values in both the "congruent" and "incongruent" values for each key separately.

I then need to use the boolean values in "correct" for each key to calculate a percentage of true in each one.

I am managing to loop through the first couple of layers on the dictionary however now I am a little lost I'm not sure how to go into the next layer down and begin making the stats calculations I need.

Hope this is clear enough for people. Let me know if you have any questions. Thanks in advance!

Answer 1

Based on the example of rts given, you can construct a dictionary with statistics with this code fragment:

import statistics
import json

rts = { ... as given ... }

stats_dict = {}
for k in rts.keys():
    stats_dict[k] = {}
    for ck in rts[k].keys():
        stats_dict[k][ck] = {}
        stats_dict[k][ck]["mean"] = statistics.mean(rts[k][ck]["rt"])
        stats_dict[k][ck]["stdev"] = statistics.stdev(rts[k][ck]["rt"])    
        stats_dict[k][ck]["true_percentage"] = len([x for x in rts[k][ck]["correct"] if x == "True"]) / len(rts[k][ck]["correct"])

print(json.dumps(stats_dict, indent=2))

Notes

• You do not necessarily need numpy to calculate the staticstis. The built-in statistics package is sufficient
• You do not need to initialize the dictionary in advance. Just make use of the keys of the given dictionary rts and use the same keys for the statisics dictionary stats_dict

Output

{
  "PO1": {
    "congruent": {
      "mean": 0.6205138,
      "stdev": 0.07207165926839758,
      "true_percentage": 0.8333333333333334
    },
    "incongruent": {
      "mean": 0.6205138,
      "stdev": 0.07207165926839758,
      "true_percentage": 0.8333333333333334
    }
  },
  "PO2": {
    "congruent": {
      "mean": 0.6205138,
      "stdev": 0.07207165926839758,
      "true_percentage": 0.8333333333333334
    },
    "incongruent": {
      "mean": 0.6205138,
      "stdev": 0.07207165926839758,
      "true_percentage": 0.8333333333333334
    }
  }
}