Numpy - Load Csv With First Row As Names Immediately To A Structured Array?

Is there a way to avoid having to predefine the names of column headers in numpy/pandas to create a structured array, and instead have numpy/pandas read in the first row as the hea

Solution 1:

For the sake of completeness, a numpy example:

rec_arr = np.genfromtxt("try.csv", delimiter = " ", names=True, dtype=None)
rec_arr
array([(b'2015-08-08',  266.,  280.04,  266.82),
       (b'2015-07-08',  233.,  280.04,  266.82)],
      dtype=[('Date', 'S10'), ('low', '<f8'), ('open', '<f8'), ('close', '<f8')])

Then, you can access columns like:

rec_arr['close']
array([ 266.82,  266.82])

and do some math as usual:

rec_arr['close'].mean()
266.81999999999999

Solution 2:

If no parameter names in read_csv then first row data create columns of df.

So work:

df = pd.read_csv('file.csv')
#if necessary change default parameter sep=','
df = pd.read_csv('file.csv', sep=';')
print (df)
         Date    low    open   close
0  2015-08-08  266.0  280.04  266.82
1  2015-07-08  233.0  280.04  266.82

print (df.columns)
Index(['Date', 'low', 'open', 'close'], dtype='object')