For loop with multiple IF conditions in Pandas

Question

I want to make for loop with the condition over columns in panda DataFrame:

import numpy as np  
import pandas as pd


df=pd.DataFrame(pd.read_csv("data.csv"))  
print df  

DWWC1980     DWWC1985   DWWC1990  
16.7140310  16.35661439 15.89201716  
20.9414479  18.00822799 15.73516051  
33.95022337 51.87065104 73.76376497  
144.7000805 136.1462017 130.9143924  
54.9506033  75.03339188 93.22994974

For loop condition statement:

for i in range (1980,2015,5):

    if   any(df["DWWC"+str(i)] <=18.25)  :

            df['MWTP'+str(i)]=(((10-33)/(5))*(df["DWWC"+str(i)]-5))+10  

    elif any((df["DWWC"+str(i)] >  18.25) &  (df["DWWC"+str(i)] <= 36.5)) :

            df['MWTP'+str(i)]=((10/(df.two-df.three))*(df["DWWC"+str(i)]-df.three))+df.Three

    else :
            df['MWTP'+str(i)]=(((df.Three_value-6)/(df.three-5))*(df["DWWC"+str(i)]-6  

df.to_csv('MWTP1.csv',index='ISO3')

But When I run this code and compare with manual calculation, I found that only the first condition calculation is correct and didn't go for the other conditions. (df.one, df.two, and df.three are other columns.)

  MWTP1980       MWTP1985         MWTP1990  
 25.87096095    30.72758886  37.04060109  
 -77.06996017   20.00112954      95.22533503  
 -290.1012655   -640.6304196    -1068.866556  
 -1845.172654   -1718.865351    -1641.61201  
 -1397.638671   -2171.737373    -2873.130596

So what is the output you are getting vs. the one you would want to get? Please provide complete MCVE. — sophros
– sophros, Commented Jan 19, 2019 at 7:42
If you want to check against all conditions, don't use elif and else; just use multiple if-statements and they will all be checked independently — Jondiedoop
– Jondiedoop, Commented Jan 19, 2019 at 7:44

jezrael · Accepted Answer · 2019-01-19 08:15:18Z

1

You can use numpy.select and for get columns names format:

for i in range (1980,2015,5):
    m1 = df["DWWC{}".format(i)] <=18.25
    #inverted m1 mask by ~
    m2 = ~m1 & (df["DWWC{}".format(i)] <= 36.5)
    a = (((10-33)/(5))*(df["DWWC{}".format(i)]-5))+10 
    b = ((10/(df.two-df.three))*(df["DWWC{}".format(i)]-df.three))+df.Three
    c = (((df.Three_value-6)/(df.three-5))*(df["DWWC{}".format(i)]-6

    df["MWTP{}".format(i)] = np.select([m1,m2],[a,b], default=c)

edited Jan 19, 2019 at 8:15

answered Jan 19, 2019 at 7:44

jezrael

868k103 gold badges1.4k silver badges1.3k bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

water77 Over a year ago

Thank you, but why in the last line you separated just "c"?

jezrael Over a year ago

@water77 it is default value, for better readability added to code.

Dor Shinar · Accepted Answer · 2019-01-19 07:46:10Z

1

I believe your problem is the usage of if elif else as follows:

if any(df["DWWC"+str(i)] <=18.25):
// executes if confidion is true
elif any((df["DWWC"+str(i)] >  18.25) &  (df["DWWC"+str(i)] <= 36.5)):
// executes if first condition is false and second condition is true
else:
// executes if both condition are false

So when your first condition is met, it never checks the other ones. Try changing it to something like that:

if any(df["DWWC"+str(i)] <=18.25):
// executes if first condition is true
if any((df["DWWC"+str(i)] >  18.25) &  (df["DWWC"+str(i)] <= 36.5)):
// executes if second condition is true, regardless of the first
else:
// all other if's are false

answered Jan 19, 2019 at 7:46

Dor Shinar

1,5222 gold badges11 silver badges17 bronze badges

4 Comments

jezrael Over a year ago

Yes, it is one problem, but solution still not working, because always all data should be overwritten.

water77 Over a year ago

@Dor Shinar thank you, Do you think a.any() cause some problems here?

water77 Over a year ago

Yes, you are right @jezrael, now the first condition calculation is wrong due to over writing.

Dor Shinar Over a year ago

any() is used to test an array of conditions to see if at least one of them is truthy. Here though it looks as if you are testing only one condition, so it really doesn't seem necessary. One more thing I've noticed is your use of the & symbols. The & symbols is used for bitwise operations, rather that logical and in python (Unlike other languages you may know). If you wanted to make an and condition, change it to if condition1 and condition2:.

Collectives™ on Stack Overflow

For loop with multiple IF conditions in Pandas

2 Answers 2

2 Comments

4 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

2 Comments

4 Comments

Your Answer

Sign up or log in

Post as a guest

Related