CODEWITHSUNDEEP
pythonpandasunpack
jcoke
jcoke

Reputation: 1891

Explode and unpack column in Pandas

Assume I have a data with this structure below, how would one explode the column that contains a list then unpack the exploded column?

Source:

d = { 
    "_id" : "5f2", 
    "connId" : 128, 
    "hospitalList" : [
        {
            "hospitalId" : 29, 
            "boardId" : 1019, 
            "siteId" : 1
        }, 
        {
            "hospitalId" : 3091, 
            "boardId" : 2163, 
            "siteId" : 382
        },
      {
            "hospitalId" : 28, 
            "boardId" : 1017, 
            "siteId" : 5
        }]
    }

Code:

root = pd.json_normalize(d)
nested_cols = [i for i in root.columns if isinstance(root[i][0], list)]
l = [root.drop(nested_cols,1),]
for i in nested_cols:
    l.append(pd.json_normalize(d, record_path=i))

output = pd.concat(l, axis=1)
print(output)

Actual Result:

                        _id      connId  hospitalId       boardId  siteId
0                       5f2       128.0          29         1019       1
1                       NaN         NaN        3091         2163     382
2                       NaN         NaN          28         1017       5

Expected Result:

                        _id      connId  hospitalId      boardId  siteId
0                       5f2       128.0          29         1019       1
1                       5f2       128.0        3091         2163     382
2                       5f2       128.0          28         1017       5

Upvotes: 0

Views: 124

Answers (1)

Alex Metsai
Alex Metsai

Reputation: 1950

This outputs what you want.

root = pd.json_normalize(d)
nested_cols = [i for i in root.columns if isinstance(root[i][0], list)]
l = [root.drop(nested_cols,1),]
for i in nested_cols:
    l.append(pd.json_normalize(d, record_path=i))

output = pd.concat(l, axis=1)

output.fillna(method='ffill', inplace=True)

Though, unfortunately, I can't know under which circumstances you're going to use the code, and/or if you'll have to make adjustments.

Upvotes: 1

Related Questions