CODEWITHSUNDEEP
pythonpandas
Doctor_W
Doctor_W

Reputation: 45

Pandas: how to test a result?

Imagine that you have a simple dataframe as such 

   low_signal  high_signal  condition  prevision
0         1.2          1.3          1        1.7
1         1.7          1.8          1        1.8
2         1.9          2.0          1        1.9
3         1.6          1.7         -1        1.5
4         1.3          1.4         -1        1.4

Now I want to study if my prediction is correct and if so I would like to know the time (which in this case we can say that is equal to the index)

My first condition

if df.condition == 1

the second is to check if df.prevision will ever be lower or equal than the df.low_signal.

If that condition is True then the 

return is the index (in study) of the df.predicted - the index of df.low_signal


   low_signal  high_signal  condition  prevision verification
0         1.2          1.3          1        1.7            1
1         1.7          1.8          1        1.8            1
2         1.9          2.0          1        1.9            0
3         1.6          1.7         -1        1.5            1
4         1.3          1.4         -1        1.4            0

That is just for df.condition == 1

For df.condition == -1 is almost the same, however the check is if the df.prevision will ever be greater or equal than the df.high_signal

Upvotes: 2

Views: 92

Answers (1)

lrnzcig
lrnzcig

Reputation: 3947

A simple way to solve it:

In [18]:

df['verification'] = ((df['high_signal'] <= df['prevision']) & (df['condition'] == 1)) \
                     | ((df['low_signal'] <= df['prevision']) & (df['condition'] == -1))

In [19]:

df
Out[19]:
            low_signal  high_signal condition   prevision   verification
0           1.2         1.3         1           1.7         True
1           1.7         1.8         1           1.8         True
2           1.9         2.0         1           1.9         False
3           1.6         1.7         -1          1.5         False
4           1.3         1.4         -1          1.4         True

Probably you know that in python True and False correspond to 1 and 0, but if you want to see the verification column as int, just do:

In [20]:

df['verification'] = df['verification'].astype(int)

Hope it helps.

EDIT: I update the answer as per your comments. I don't really understand the 'functional requirement', but anyway if you have a more complex operation you can use pandas' apply

In your case is a bit tricky since:

  • you want to access the index inside the apply function, and I don't think you can
  • you want to iterate over the whole dataframe

But anyway, solving the expression you've put in your comment, you can first add a column with a copy of the index in your dataframe,

In [27]:

df['index_cp'] = df.index

and then use the apply function,

In [28]:

len_df = len(df)
​
def f(x):
    for n in range(len_df):    
        if x.condition == 1 & (x.prevision <= df.low_signal[n]):
            return n - x.index_cp
​
df.apply(f, axis=1)
Out[28]:
0     1
1     1
2     0
3   NaN
4   NaN
dtype: float64

Probably you need to modify the function but as I say I did not get the "functional" view and I've just implemented what you've posted. Hope it helps.

Upvotes: 2

Related Questions