I wanted to save variables all at once, so I made such a program. It can be saved by applying a two-dimensional array.

... I noticed after writing, but the method of adding may be better [If you are busy, jump to the postscript](# postscript) If you simply want to save all variables, use dill

With the append method, an error occurred in the type relationship or the data you tried to use. I didn't get an error when using DataFrame, so It may not be meaningless to insert a DataFrame once.

import pandas as pd

First import pandas


# ---Define variables for testing---
a ,b,c,d = 1,2,3,4
xx = [3,6,8]
yy = [5,8,2]
zz = [8,2,8]

Save this function, array

# ---Collect variables into dataframe and save csv---
def getName(obj):
    return  [k for k, v in globals().items() if id(obj) == id(v)][0] #Return variable name with str

Get the variable name as str by using this function It is almost a copy of this article Click here for details

df = pd.DataFrame()

datas1 = [xx,yy,zz] #While using the variable name as a header, put together a one-dimensional array into a dataframe
for data in datas1:
    df = pd.concat([df, pd.DataFrame({getName(data):data})],axis=1) # {}Add dataframe with concat using set

Create a DataFrame based on {name: value} and save more and more to df with concat. By inserting axis = 1, it will be added horizontally

datas2 = [a,b,c,d]#While using the variable name as a header, group the variables into a dataframe
for data in datas2:
    df =  pd.concat([df, pd.DataFrame({getName(data):[data]})],axis=1)# {}Add dataframe with concat using set

Do the same thing as before The value of{name: value}must be an array, so the value is [data]

df.to_csv('test.csv')
print(df)
del df

Save and display. Since df tends to be used in various places, I will delete it for the time being.

You can see that it is displayed and saved properly in the table below.

   xx  yy  zz    a    b    c    d
0   3   5   8  1.0  2.0  3.0  4.0
1   6   8   2  NaN  NaN  NaN  NaN
2   8   2   8  NaN  NaN  NaN  NaN

Source code

import pandas as pd

# ---Define variables for testing---
a ,b,c,d = 1,2,3,4
xx = [3,6,8]
yy = [5,8,2]
zz = [8,2,8]

# ---Collect variables into dataframe and save csv---
def getName(obj):
    return  [k for k, v in globals().items() if id(obj) == id(v)][0] #Return variable name with str

df = pd.DataFrame()

datas1 = [xx,yy,zz] #While using the variable name as a header, put together a one-dimensional array into a dataframe
for data in datas1:
    df = pd.concat([df, pd.DataFrame({getName(data):data})],axis=1) # {}Add dataframe with concat using set
    
datas2 = [a,b,c,d]#While using the variable name as a header, group the variables into a dataframe
for data in datas2:
    df =  pd.concat([df, pd.DataFrame({getName(data):[data]})],axis=1)# {}Add dataframe with concat using set
df.to_csv('test.csv')
print(df)
del df

Postscript

Write directly to csv

# ---Define variables for testing---
a ,b,c,d = 1,2,3,4
xx = [3,6,8]
yy = [5,8,2]
zz = [8,2,8]

import csv
# ---Select a variable and save csv---
def getName(obj):
    return  [k for k, v in globals().items() if id(obj) == id(v)][0] #Return variable name with str

with open('test2.csv','a', newline="") as f:
    datas = [xx,yy,zz,]
    writer = csv.writer(f)
    for data in datas:
        writer.writerow([getName(data)]+data)
    datas = [a,b,c,d]
    for data in datas:
        writer.writerow([getName(data)]+[data])

Case division is to convert a single variable to list type

Output test2.csv

xx,3,6,8
yy,5,8,2
zz,8,2,8
a,1
b,2
c,3
d,4

How to use dill

Save


import dill
dill.dump_session('session.pkl')

Read

import dill
dill.load_session('session.pkl')

All variables can be used as they are.

[python] Use DataFrame to label arbitrary variables and arrays together and save them in csv [pandas]

Postscript

How to use dill