CODEWITHSUNDEEP
pythonpandasfeature-engineeringtechnical-indicator
alitkie
alitkie

Reputation: 13

create a feature in pandas based on another column (i.e. when it goes from positive to negative)

I have the following dataset:

data = {'ROC_9': [0.006250, 0.087230, 0.045028, 0.165738, -0.006993, -0.432736, -0.11162, 0.057466, 0.203138, -0.008234]}

price_data = pd.DataFrame (data)

It is an indicator about a stock price, namely rate of change.

I want to write a code that creates a new feature (column) on the pandas data frame when a current feature on the pandas data frame goes from positive to negative, or vica versa.

It is easier explained through an example: lets use the feature ROC9.

I create a new variable called ROC9_signal and set it equal to 0:

price_data['ROC9_signal'] = 0

When ROC_9 goes from negative to positive, I want to change the ROC9_signal from 0 to 1.

When ROC_9 goes from positive to negative, I want to change the ROC9_signal from 0 to -1.

Looking at the data, I would like ROC9_signal to change from 0 to -1, since the value has gone from 0.16 (positive) to -0.006 (negative).

Looking at the data, I would like ROC_9 signal to change from 0 to 1, since the value has gone from -0.11 (negative) to 0.05 (positive).

Looking at the data, I would like ROC9_signal to change from 0 to -1, since the value has gone from 0.20 (positive) to -0.008 (negative).

It is only the row where the change happens that I want to change from 0 to 1 or 0 to -1, the other rows must remain at 0.

I will then apply this same logic to create a momentum10_signal column and a chalkin_money_flow_signal column. Therefore I want a solution that can be applied to different columns and not manually.

Thanks in advance for the help.

This is what the full data looks like:

Full Data

Upvotes: 0

Views: 170

Answers (1)

Quang Hoang
Quang Hoang

Reputation: 150785

You can use np.sign to extract the signs. Something like this:

signs = np.sign(price_data.ROC_9)
price_data['signal'] = np.sign(signs.diff()).fillna(0)

Output:

      ROC_9  signal
0  0.006250     0.0
1  0.087230     0.0
2  0.045028     0.0
3  0.165738     0.0
4 -0.006993    -1.0
5 -0.432736     0.0
6 -0.111620     0.0
7  0.057466     1.0
8  0.203138     0.0
9 -0.008234    -1.0

Upvotes: 1

Related Questions