CODEWITHSUNDEEP
pythonpandasdataframe
Behdad
Behdad

Reputation: 1629

Comparing current row and previous row in Pandas

I have a stock dataframe:

   Open     High      Low    Close  Volume        rb          us  \
0  1.20821  1.20821  1.20793  1.20794  138.96  0.022347  100.000000   
1  1.20794  1.20795  1.20787  1.20788  119.61  0.004967   85.714286   
2  1.20788  1.20793  1.20770  1.20779  210.42  0.007451   64.285714   
3  1.20779  1.20791  1.20779  1.20789   77.51  0.008280   83.333333   
4  1.20789  1.20795  1.20789  1.20792   56.97  0.002484   50.000000   

           ls  color  
0   96.428571  black  
1   85.714286  black  
2   50.000000  black  
3  100.000000  white  
4  100.000000  white

I want to compare current row data to previous row data,something like this:

if(df['color'] == df['color'].shift(-1)):
   if((df['Close'] >= df['Open'].shift(-1) and df['Open']>=df['Close'].shift(-1)):
      df['Position'] = UP
   if((df['Close'] < df['Open'].shift(-1) and df['Open']<=df['Close'].shift(-1)):
      df['Position'] = DOWN

THERE ARE MANY MORE IF CONDITIONS ...

It's not possible to compare data by 

np.where(condition,TRUE,FALSE)

Because of many conditions I have in my algorithm. It's just part of them.

How can I do these conditions check?

Upvotes: 2

Views: 12209

Answers (2)

Anton vBR
Anton vBR

Reputation: 18916

I would divide it in different check-ups:

diffdf = df.diff()[1:]

cond1 = diffdf['Volume'] > 0
cond2 = diffdf['Close'] >= 0
# cond3 = diffdf['Open'] >= 0 ?? What is this supposed to check?

df['Position'] = np.insert(np.where(cond1&cond2, 'UP', 'DOWN'), 0, '-')

df.diff():

      Open     High      Low    Close  Volume
0      NaN      NaN      NaN      NaN     NaN
1 -0.00027 -0.00026 -0.00006 -0.00006  -19.35
2 -0.00006 -0.00002 -0.00017 -0.00009   90.81
3 -0.00009 -0.00002  0.00009  0.00010 -132.91
4  0.00010  0.00004  0.00010  0.00003  -20.54

Full example:

import pandas as pd
import numpy as np

data = '''\
Open     High      Low    Close  Volume
1.20821  1.20821  1.20793  1.20794  138.96
1.20794  1.20795  1.20787  1.20788  119.61
1.20788  1.20793  1.20770  1.20779  210.42
1.20779  1.20791  1.20779  1.20789   77.51
1.20789  1.20795  1.20789  1.20792   56.97'''

fileobj = pd.compat.StringIO(data)
df = pd.read_csv(fileobj, sep='\s+')

# Our client wants to know what stocks have increased in Volumne and Close
# We call this element Position and it is either Up,Down,-
# Let us create a difference dataframe and check those conditions
diffdf = df.diff()[1:]
cond1 = diffdf['Volume'] > 0
cond2 = diffdf['Close'] >= 0
df['Position'] = np.insert(np.where(cond1&cond2, 'UP', 'DOWN'), 0, '-')

print(df)

Returns:

      Open     High      Low    Close  Volume Position
0  1.20821  1.20821  1.20793  1.20794  138.96        -
1  1.20794  1.20795  1.20787  1.20788  119.61     DOWN
2  1.20788  1.20793  1.20770  1.20779  210.42     DOWN
3  1.20779  1.20791  1.20779  1.20789   77.51     DOWN
4  1.20789  1.20795  1.20789  1.20792   56.97     DOWN

Upvotes: 2

harpan
harpan

Reputation: 8631

You need:

df['Position'] = np.where((df['Volume'] > df['Volume'].shift(-1)) & ((df['Close'] >= df['Close'].shift(-1)) & (df['Open'] <= df['Open'])),"UP","DOWN")

Output:

        Open    High    Low Close   Volume  Position
0   1.20821 1.20821 1.20793 1.20794 138.96  UP
1   1.20794 1.20795 1.20787 1.20788 119.61  DOWN
2   1.20788 1.20793 1.20770 1.20779 210.42  DOWN
3   1.20779 1.20791 1.20779 1.20789 77.51   DOWN
4   1.20789 1.20795 1.20789 1.20792 56.97   DOWN

Upvotes: 8

Related Questions