Creating dataframe from dictionary in pyspark

failure.py

# Ran following examples in PySpark shell 
from pyspark.sql import Row

nested_struct = {
    "org_id": "abc", "emp_id": "def",
    "projects": [{"name": "test_proj_1", "duration": 20, "code": "p30"},
                 {"name": "test_proj_2", "duration": 15, "code": "p30"}]
}
test_rdd = sc.parallelize([Row(**nested_struct)])
test_df = test_rdd.toDF()

print test_rdd.collect()[0].projects
# Output: [{'duration': 20, 'code': 'p30', 'name': 'test_proj_1'}, {'duration': 15, 'code': 'p30', 'name': 'test_proj_2'}]
print test_df.collect()[0].projects
# Output: [{u'duration': 20, u'code': None, u'name': None}, {u'duration': 15, u'code': None, u'name': None}]

success.py

# Ran following in PySpark shell
from pyspark.sql import Row

nested_struct = {
    "org_id": "abc", "emp_id": "def",
    "projects": [{"name": "test_proj_1", "duration": "20", "code": "p30"},
                 {"name": "test_proj_2", "duration": "15", "code": "p30"}]
}
test_rdd = sc.parallelize([Row(**nested_struct)])
test_df = test_rdd.toDF()

print test_rdd.collect()[0].projects
# Output: [{'duration': '20', 'code': 'p30', 'name': 'test_proj_1'}, {'duration': '15', 'code': 'p30', 'name': 'test_proj_2'}]
print test_df.collect()[0].projects
# Output: [{u'duration': u'20', u'code': u'p30', u'name': u'test_proj_1'}, {u'duration': u'15', u'code': u'p30', u'name': u'test_proj_2'}]