CODEWITHSUNDEEP
pythonpandasnumpydataframe
Hacen Selahy
Hacen Selahy

Reputation: 79

dataframe type lost after saving

I have the Dataframe I changed to category type. but after saving and loading the type returned to float64 

data = pd.DataFrame(barray, columns=['line 1', 'line 2', 'line 3', 'line 4', 'piece to move', 'place to occupy'])
    data['possibility of win'] = y
    for col in ['line 1', 'line 2', 'line 3', 'line 4', 'piece to move', 'place to occupy']:
        data[col] = data[col].astype('category')

    data['line 1'].dtype

output: CategoricalDtype(categories=['2000', '2001', '2002', '2010', '2011', '2012', '2020', '2021', '2100', '2101', '2102', '2110', '2112', '2120', '2121', '2200', '2201', '2202', '2210', '2211', '2212', '2220', '2221'], ordered=False)

data.to_csv('data.csv')
data = pd.read_csv("data.csv")
data['line 1'].dtype

output: dtype('int64')

Upvotes: 1

Views: 279

Answers (1)

jezrael
jezrael

Reputation: 863166

It is expected, because in csv all data are saved like text.

Solution is add parameter dtype in read_csv:

#all columns set to categoricals
data = pd.read_csv("data.csv", dtype='category')

If want specify only some columns use dictionary:

cols = ['line 1', 'line 2', 'line 3', 'line 4']
d = dict.fromkeys(cols, 'category')
data = pd.read_csv("data.csv", dtype=d)

Upvotes: 1

Related Questions