Two dataframes train and test. The train dataframe has columns A, Z, D, C and test has columns C, D, Z. How do I take train's columns and make test dataframe also has four columns with the same order as train? The newly added columns A should have value 0 for all rows.
Thanks.
You can use assign to create a new column A on test and use list of column names to reorder the columns:
test.assign(A = 0)[['A', 'Z', 'D', 'C']]
Or: test.assign(A = 0)[train.columns]
test = pd.DataFrame({
'C': [1,2,3],
'D': [2,3,4],
'Z': [3,4,5]
})
test.assign(A = 0)[['A','Z','D','C']]
# A Z D C
#0 0 3 2 1
#1 0 4 3 2
#2 0 5 4 3
Or using reindex
test['A']=0
test.reindex_axis(train.columns, axis=1)
It is also possible to use pd.concat as follows.
import pandas as pd
import numpy as np
df_train = pd.DataFrame({
"A": np.random.randn(10),
"Z": np.random.randn(10),
"D": np.random.randn(10),
"C": np.random.randn(10)
})
df_test = pd.DataFrame({
"Z": np.random.randn(10),
"D": np.random.randn(10),
"C": np.random.randn(10)
})
df_all = pd.concat([df_train, df_test], axis=0)
df_all = df_all.fillna(0)
df_train = df_all.iloc[0:len(df_train.index)]
df_test = df_all.iloc[len(df_train.index)+1:len(df_all.index)]
