How to create dictionary from multiple dataframes?

I have a folder with several csv files. Example of the dataframes from csv files in directory:

d1 = {'timestamp': [2013-01-30, 2015-02-29, 2014-03-25, 2016-01-01, 2018-02-20, 
2012-05-05, 2018-02-04], 
     'site': ['plus.google.com','vk.com','yandex.ru','plus.google.com','vk.com', 'oracle.com', 'oracle.com']}
df1 = pd.DataFrame(data = d)

  
d2 = {'timestamp': [2013-01-30, 2015-02-29, 2014-03-25, 2016-01-01, 2018-02-20,], 
             'site': ['plus.google.com','meduza.ru','yandex.ru','google.com', 'meduza.ru'}
df2 = pd.DataFrame(data = d2)

I need to make a function that accepts route to file directory and return sites frequency dictionary (one for all sites in file directory) with unique sites names the following kind: {‘site_string’: [site_id, site_freq]} For our examle it will be: {‘vk.com’: (1, 2), ‘plus.google.com’: (2, 3), ‘yandex.ru’: (3, 2), ‘meduza.ru’: (4, 2), ‘oracle.com’: (5, 2), ‘google.com’: (6, 1)}

I tried to aply value_counts() to every dataframe, made dictionaries of them and tried to concatenate dicts, but duplicates are removed in this case. How I can solve this issue? What should I do?

 def prepare_train_set(path_to_csv_files):
        frequency = {}
        for filename in glob(f'{path_to_csv_files}/*'):
            sub_iterationed_df = pd.read_csv(filename)
            value_counts_dict = dict(sub_iterationed_df["site"].value_counts())
            frequency.update(value_counts_dict)
        return frequency

Also I tried to make lists from keys and values of value_counts() dictionary and after that make dictionary with zip function but there is an error “list assignment index out of range”. WHy this error occurs and how can I bypass the error?

def CheckForDuplicates(keys_list, values_list):
    keys_list = list(value_counts_dict.keys())
    values_list = list(value_counts_dict.values())
        keys_list_constant = keys_list[:]
        values_list_constant = values_list[:]
        for i in range(len(keys_list_constant)):
            checking_dup_keys_list = keys_list[:i]
            checking_dup_values_list = values_list[:i]
            key_value = keys_list_constant[i]
            if  key_value in checking_dup_keys_list:
                duplicate_index = checking_dup_keys_list.index(key_value)
                values_list[duplicate_index] = values_list[duplicate_index] + values_list_constant[i]
                del values_list[i]
                del keys_list[i]
            return(keys_list, values_list)
CheckForDuplicates(keys_list, values_list)

Answer

You could use a Counter instead of a plain dictionary:

from collections import Counter

def prepare_train_set(path_to_csv_files):
    frequency = Counter()
    for filename in glob(f'{path_to_csv_files}/*'):
        sub_iterationed_df = pd.read_csv(filename)
        value_counts_dict = sub_iterationed_df['site'].value_counts().to_dict()
        frequency.update(value_counts_dict)
    return frequency

From the docs:

update([iterable-or-mapping]):
Elements are counted from an iterable or added-in from another mapping (or counter). Like dict.update() but adds counts instead of replacing them.

Or concatenate all the dataframes and take the .value_counts() afterwards.

Advertisement

Answer