I have a dictionary with thousand keys something as follows:
my_dictionary:
{'key1':[['ft1',[2,4,12,2]],['ft2',[0,3,3,1]],'key2':[['ft1',[5,0,2,9]],['ft2',[10,39,3,2]]}
Now, I want to convert this dictionary into a dataframe, where the keys should be a specific column, and the features (ft1, ft2, ..) and their values converted to different column as well. So my desired dataframe should be like this:
my_new_dataframe:
ID, ft1_sig,ft1_med,ft1_les,ft1_non,ft2_sig,ft2_med,ft2_les,ft2_non,...
key1 2 4 12 2 0 3 3. 1
key2 5. 0. 2. 9. 10. 39 3. 2
...
keyn. .. .. .. ..
I attempted a solution at this, but it requires that each of the keys (ie key1, key2, etc.) contains the ft attribute you want in a dictionary. Also, are you missing a "]" for the original list? It was mismatched when I pasted into my interpreter.
import pandas as pd
#added method to change your original dictionary to one that I can manipulate with the method below.
#If you compare the values of new_dict and data using ==, it returns true.
my_dictionary = {'key1':[['ft1',[2,4,12,2]],['ft2',[0,3,3,1]]],'key2':[['ft1',[5,0,2,9]],['ft2',[10,39,3,2]]]}
new_dict ={}
for element in my_dictionary:
print(element)
print(my_dictionary[element])
new_dict[element] = dict(my_dictionary[element])
print(new_dict)
data = {
'key1':{
'ft1':[2,4,12,2],
'ft2':[0,3,3,1]
},
'key2':{
'ft1':[5,0,2,9],
'ft2':[10,39,3,2]
}
}
keys = list(data.keys())
df = pd.DataFrame.from_dict(data).T
df2 = pd.DataFrame(df.ft1.values.tolist()).add_prefix('ft1_')
df3 = pd.DataFrame(df.ft2.values.tolist()).add_prefix('ft2_')
df4 = pd.merge(df2,df3,left_index=True,right_index=True)
df4.index=keys
print(df4)
Here is the output:
I added more data in your example to show that the script will be flexible if new features or rows (keys) are added.
Here it is
mydict = {'key1':[['ft1',[2,4,12,2]],['ft2',[0,3,3,1]],['ft3',[0,3,3,1]]],
'key2':[['ft1',[5,0,2,9]],['ft2',[10,39,3,2]],['ft3',[0,3,3,1]]]
,'key3':[['ft1',[5,0,2,9]],['ft2',[10,39,3,2]],['ft3',[0,3,3,1]]]}
df = pd.DataFrame(mydict).T
colname = [df[c][0][0] for c in df]
df = df.applymap(lambda c: c[1])
df.reset_index(level=0, inplace=True)
df.columns=['ID'] + colname
s=['_sig','_med','_les','_non']
def f(x):
return pd.Series([x[0], x[1], x[2], x[3]])
for col in colname:
df[[col+'_sig', col+'_med', col+'_les', col+'_non']]= df[col].apply(lambda x: f(x))
df.drop(colname, axis=1, inplace=True)
df
Result:
ID ft1_sig ft1_med ft1_les ft1_non ft2_sig ft2_med ft2_les ft2_non ft3_sig ft3_med ft3_les ft3_non
0 key1 2 4 12 2 0 3 3 1 0 3 3 1
1 key2 5 0 2 9 10 39 3 2 0 3 3 1
2 key3 5 0 2 9 10 39 3 2 0 3 3 1
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.