pandas.DataFrame.interpolate — pandas 0.16.2 documentation (original) (raw)

method : {‘linear’, ‘time’, ‘index’, ‘values’, ‘nearest’, ‘zero’,

‘slinear’, ‘quadratic’, ‘cubic’, ‘barycentric’, ‘krogh’, ‘polynomial’, ‘spline’ ‘piecewise_polynomial’, ‘pchip’}

• ‘linear’: ignore the index and treat the values as equally spaced. default
• ‘time’: interpolation works on daily and higher resolution data to interpolate given length of interval
• ‘index’, ‘values’: use the actual numerical values of the index
• ‘nearest’, ‘zero’, ‘slinear’, ‘quadratic’, ‘cubic’, ‘barycentric’, ‘polynomial’ is passed toscipy.interpolate.interp1d with the order given both ‘polynomial’ and ‘spline’ requre that you also specify and order (int) e.g. df.interpolate(method=’polynomial’, order=4)
• ‘krogh’, ‘piecewise_polynomial’, ‘spline’, and ‘pchip’ are all wrappers around the scipy interpolation methods of similar names. See the scipy documentation for more on their behavior:http://docs.scipy.org/doc/scipy/reference/interpolate.html#univariate-interpolation http://docs.scipy.org/doc/scipy/reference/tutorial/interpolate.html

axis : {0, 1}, default 0

• 0: fill column-by-column
• 1: fill row-by-row

limit : int, default None.

Maximum number of consecutive NaNs to fill.

inplace : bool, default False

Update the NDFrame in place if possible.

downcast : optional, ‘infer’ or None, defaults to None

Downcast dtypes if possible.