CODEWITHSUNDEEP
pythonpandas
Igor K.
Igor K.

Reputation: 877

Pandas: How to find values that cross zero?

This is my DataFrame

      Date     Time      Value
16.02.2020 21:00:00    0.05012
16.02.2020 22:00:00    0.04285
16.02.2020 23:00:00    0.03559
17.02.2020  0:00:00    0.02833
17.02.2020  1:00:00    0.02107
17.02.2020  2:00:00    0.01380
17.02.2020  3:00:00    0.00654
17.02.2020  4:00:00   -0.00073
17.02.2020  5:00:00   -0.00799
17.02.2020  6:00:00   -0.01526
17.02.2020  7:00:00   -0.02252
17.02.2020  8:00:00   -0.02978
17.02.2020  9:00:00   -0.03704
17.02.2020 10:00:00   -0.04430
17.02.2020 11:00:00   -0.05156

How to find rows where Value crosses Zero? I can do it with iterating:

if current value > 0 and previous value < 0,
or
current value < 0 and previous value > 0,

so it is zero-crossing.
But my dataframe is very large. How to do it without iterating?

Upvotes: 4

Views: 1314

Answers (2)

Terry
Terry

Reputation: 2811

we can create masks to make this filter with shift

mask1 = (df['Value'] < 0)
mask2 = (df['Value'] > 0).shift()

mask3 = (df['Value'] > 0)
mask4 = (df['Value'] < 0).shift()

df.loc[(mask1 & mask2) | (mask3 & mask4)]
#output:
    Date        Time    Value
7   17.02.2020  4:00:00 -0.00073

Upvotes: 2

anky
anky

Reputation: 75080

IIUC, you can try np.sign + series.diff

out = df[np.sign(df['Value']).diff().fillna(0).ne(0)].copy()

Upvotes: 4

Related Questions