Converting pandas dataframe to dictionary with same keys over multiple rows

user_16799738 0
Viewed 384

I'm trying to create a dictionary from a pandas data frame with key from a column and values from rest of columns. But the problem is I will have same keys in multiple rows and I've read through many other similar SO posts but couldn't find the answer. This is what I have:

df1:

pid  feature_id  feature_value
78          20            1.0
78        1130            3.0
...
91        1148            1.0
92        1153            4.0
92        1154            1.0
...
115       1162            1.0
115       1175            5.0
......

This is what I tried:

df2 = df1.set_index('pid').agg(tuple, 1).to_dict()

But problem is this seems to not take into account the same keys from multiple rows.

What I want is something like this:

{78: [(20, 1.0), (1130, 3.0)]..., 115: [(1162, 1.0), (1175, 5.0)], ...}

Please advise.

pandas dataframe dictionary python
2 Answers

Take the example dataframe:

df = pd.DataFrame({'col0':[1,2,3,1,2],'col1':[10,20,30,40,50],'col2':[7,8,9,8,7]})

You can do:

df = df.assign(pairs = df.apply(lambda row: [row['col1'],row['col2']],axis=1))
res = df.groupby('col0')['pairs'].apply(list).to_dict()

Original df:

   col0  col1  col2
0     1    10     7
1     2    20     8
2     3    30     9
3     1    40     8
4     2    50     7

res:

{1: [[10, 7], [40, 8]], 2: [[20, 8], [50, 7]], 3: [[30, 9]]}

Same is applicable to your df, just replace col0, col1 and col2 by pid, feature_id and feature_value respectively.

user_85654380 
def df_to_dict(df):
    # create a dictionary
    d = {}
    # iterate over the rows
    for index, row in df.iterrows():
        # if the key is not in the dictionary, add it
        if row[0] not in d:
            d[int(row[0])] = []
        # add the tuple (row[1], row[2]) to the list associated with the key
        d[row[0]].append((row[1], row[2]))
    return d
print(df_to_dict(df))
user_102733080
Related