Convert pandas DataFrame to JSON dictionary of lists

Question

I have a pandas dataframe representing some basic weather info which looks like this

location	dDate	min-temp	max-temp	rain
Sydney	2013-01-01	15	35	10
Sydney	2013-01-02	16	36	5
Sydney	2013-01-03	16	31	0
Sydney	2013-01-04	16	31	2

To graph it using a JavaScript library such as C3, I need each column as an array in the JSON

{
  "x": ["2013-01-01", "2013-01-02", "2013-01-03", "2013-01-04"],
  "rain": [10, 5, 0, 2],
  "max-temp": [35, 36, 31, 31]
}

formatting as column doesn't quite work

df = client.query(query).to_dataframe()
jsonstr = df.to_json(orient='columns')

It results in this

{
   “location”:{
      “0”:Sydney,
      “1”:Sydney,
      “2”:Sydney,
      “3”:Sydney
   },
   "dDate":{
      "0":2013-01-01,
      "1":2013-01-02,
      "2":2013-01-03,
      "3":2013-01-04
   },
   “min-temp”:{
      “0”:15,
      “1”:16,
      “2”:16,
      “3”:16
    },
    …….
}

How can I reproduce the desired JSON where each colum is just an array of values?

Answer 1

to_json only supports records and nested dict formats. To get a dict of lists, convert to dictionary first using to_dict, then convert to json:

pd.io.json.dumps(df.to_dict(orient='list'))
# '{"x":["2013-01-01","2013-01-02","2013-01-03","2013-01-04"],"rain":[10,5,0,2],"max-temp":[35,36,31,31]}'