return rows in a dataframe closest to a user-defined number

Question

I have a user defined number which I want to compare to a certain column of a dataframe.

I would like to return the rows of a dataframe which contain (in a certain column of df, say, df.num) the 5 closest numbers to the given number x.

Any suggestions for the best way to do this without loops would be greatly appreciated.

Answer 1

I think you can use the argsort method:

>>> df = pd.DataFrame({"A": 1e4*np.arange(100), "num": np.random.random(100)})
>>> x = 0.75
>>> df.ix[(df.num-x).abs().argsort()[:5]]
         A       num
66  660000  0.748261
92  920000  0.754911
59  590000  0.764449
27  270000  0.765633
82  820000  0.732601
>>> x = 0.33
>>> df.ix[(df.num-x).abs().argsort()[:5]]
         A       num
37  370000  0.327928
76  760000  0.327921
8    80000  0.326528
17  170000  0.334702
96  960000  0.324516