HOWTO · Pandas
How to Convert Pandas Series to DataFrame
It introduces how to convert pandas series to dataframe
On this page
Use pandas.DataFrame() or pandas.Series.to_frame() to convert one Pandas Series into a DataFrame. Both methods retain the Series index as the DataFrame row index. Use to_frame(name=...) to name its value column; use reset_index(name=...) when the old index should become a regular column. To combine several Series as columns, use pd.concat(..., axis=1).
The examples use this Series, whose labels become the row index in each conversion:
import pandas as pd
import numpy as np
np.random.seed(0)
df_series = pd.Series(
np.random.randint(0, 100, size=(10)),
index=["a", "b", "c", "d", "e", "f", "g", "h", "i", "j"],
)
print(df_series)
a 44
b 47
c 64
d 67
e 67
f 9
g 83
h 21
i 36
j 87
dtype: int64
Convert a Single Pandas Series to DataFrame Using pandas.DataFrame()
Pass a Series to pandas.DataFrame() to create a one-column DataFrame. The columns argument sets the value column’s label; the Series index remains the row index. This general constructor is useful when you are already building a DataFrame from other data.
import pandas as pd
import numpy as np
np.random.seed(0)
df_series = pd.Series(
np.random.randint(0, 100, size=(10)),
index=["a", "b", "c", "d", "e", "f", "g", "h", "i", "j"],
)
print(pd.DataFrame(df_series, columns=["A"]))
A
a 44
b 47
c 64
d 67
e 67
f 9
g 83
h 21
i 36
j 87
The resulting DataFrame has one column named A and keeps the labels a through j as its index.
Use the constructor when the input is already a Series and you want to specify DataFrame options at creation. The columns list controls the output column label; it does not replace the Series index. If you omit columns, pandas uses the Series name for the value column when one is set, and otherwise uses 0.
Convert a Single Pandas Series to DataFrame Using pandas.Series.to_frame()
Call Series.to_frame() when you want to turn a Series directly into a one-column DataFrame. Pass name to choose the value column’s label. The method keeps the Series index, which is useful when those labels identify the rows.
For a single Series, to_frame(name="A") is a concise alternative to pandas.DataFrame(series, columns=["A"]). Choose a name that describes the values, especially before joining this DataFrame with other tables. These conversions do not turn the Series index into a data column; use reset_index() for that layout.
import pandas as pd
import numpy as np
np.random.seed(0)
df_series = pd.Series(
np.random.randint(0, 100, size=(10)),
index=["a", "b", "c", "d", "e", "f", "g", "h", "i", "j"],
)
print(df_series.to_frame(name="A"))
A
a 44
b 47
c 64
d 67
e 67
f 9
g 83
h 21
i 36
j 87
If a Series has no name, to_frame() uses 0 as the value column label. When you also need the Series index as data in a column, use the reset_index() method. Naming the index axis makes the first output column index:
import pandas as pd
import numpy as np
np.random.seed(0)
df_series = pd.Series(
np.random.randint(0, 100, size=(10)),
index=["a", "b", "c", "d", "e", "f", "g", "h", "i", "j"],
).rename_axis("index")
print(df_series)
print("\n")
print(df_series.reset_index())
index
a 44
b 47
c 64
d 67
e 67
f 9
g 83
h 21
i 36
j 87
dtype: int64
index 0
0 a 44
1 b 47
2 c 64
3 d 67
4 e 67
5 f 9
6 g 83
7 h 21
8 i 36
9 j 87
reset_index() returns a DataFrame with the former index in the index column and the Series values in a column named 0. Supply name to label the value column directly:
import pandas as pd
import numpy as np
np.random.seed(0)
df_series = pd.Series(
np.random.randint(0, 100, size=(10)),
index=["a", "b", "c", "d", "e", "f", "g", "h", "i", "j"],
).rename_axis("index")
print(df_series)
print("\n")
print(df_series.reset_index(name="A"))
index
a 44
b 47
c 64
d 67
e 67
f 9
g 83
h 21
i 36
j 87
dtype: int64
index A
0 a 44
1 b 47
2 c 64
3 d 67
4 e 67
5 f 9
6 g 83
7 h 21
8 i 36
9 j 87
Convert Multiple Pandas Series to Dataframes
To place multiple Series side by side in one DataFrame, pass them to pd.concat() with axis=1. Pandas aligns values by index label, so matching labels share a row. The Series below have the same index and produce three value columns:
import pandas as pd
import numpy as np
np.random.seed(0)
df_series1 = pd.Series(
np.random.randint(0, 100, size=(10)),
index=["a", "b", "c", "d", "e", "f", "g", "h", "i", "j"],
)
df_series2 = pd.Series(
np.random.randint(40, 100, size=(10)),
index=["a", "b", "c", "d", "e", "f", "g", "h", "i", "j"],
)
df_series3 = pd.Series(
np.random.randint(80, 100, size=(10)),
index=["a", "b", "c", "d", "e", "f", "g", "h", "i", "j"],
)
df_stitched = pd.concat([df_series1, df_series2, df_series3], axis=1)
print(df_stitched)
0 1 2
a 44 46 97
b 47 64 85
c 64 64 93
d 67 52 88
e 67 98 89
f 9 41 99
g 83 78 96
h 21 79 99
i 36 63 85
j 87 86 95
The pd.concat() function uses axis=1 to concatenate along columns. Its default is axis=0, which concatenates along rows instead. When indexes differ, column-wise concatenation aligns on their labels; labels absent from a Series have missing values in that column.