How to cbind (concat) 3 dataframes in Python/pandas-CodePudding

I have a dataframe

import pandas as pd

iris=pd.read_csv("https://gist.githubusercontent.com/netj/8836201/raw/6f9306ad21398ea43cba4f7d537619d0e07d5ae3/iris.csv")
iris.tail(5)
iris.head(5)

From iris dataframe I derived df_setosa,df_virginica, and df_versicolor dataframes

df_setosa = iris[iris['variety'] == 'Setosa']
df_virginica = iris[iris['variety'] == 'Virginica']
df_versicolor = iris[iris['variety'] == 'Versicolor']

# paste the corresponding variety name as the suffix to each dataframe 
df_setosa = df_setosa.add_suffix('_setosa')
df_virginica = df_virginica.add_suffix('_virginica')
df_versicolor = df_versicolor.add_suffix('_versicolor')

print(df_virginica.columns)
print(df_versicolor.columns)
print(df_setosa.columns)

print(df_setosa.shape) #  50 row by 5 columns
print(df_versicolor.shape) # 50 rows by 5 columns
print(df_virginica.shape) # 50 rows by 5 columns

Since each dataframe has shape of (50,5), I want to concatenate (or as we say in R cbind) the three dataframes.

My attempt:

#### I need help concatenating the three dataframes
concat_df  = pd.concat([df_setosa,df_virginica,df_versicolor]) # this returns a lot of NaN
concat_df.shape # this returns a shape of 150 rows by 15 columns  instead of 50 rows by 15 columns

The concat_df should have a 50 rows by 15 columns shape

Thanks in advance

CodePudding user response：

When you create the "sub" dataframes, reset their indexes, since there's no reason to keep the index of the original iris set in this case

df_setosa = iris[iris['variety'] == 'Setosa'].reset_index(drop=True)
df_virginica = iris[iris['variety'] == 'Virginica'].reset_index(drop=True)
df_versicolor = iris[iris['variety'] == 'Versicolor'].reset_index(drop=True)

Then when you concat, make sure you concat horizontally by setting "axis" argument to 1, like so:

concat_df  = pd.concat([df_setosa,df_virginica,df_versicolor], axis=1)

You can also leave the "reset_index" for this last step. If you don't do this the concat will still place 150 rows since it will try to put the indexes from 0 to 149 in order and fill the rest with NaNs