You can use the following basic syntax to import a CSV file into pandas when there are a different number of columns per row:
df = pd.read_csv('uneven_data.csv', header=None, names=range(4))
The value inside the range() function should be the number of columns in the row with the max number of columns.
The following example shows how to use this syntax in practice.
Example: Import CSV into Pandas with Different Number of Columns per Row
Suppose we have the following CSV file called uneven_data.csv:
Notice that each row does not have the same number of columns.
If we attempt to use the read_csv() function to import this CSV file into a pandas DataFrame, we’ll receive an error:
import pandas as pd #attempt to import CSV file with differing number of columns per row df = pd.read_csv('uneven_data.csv', header=None) ParserError: Error tokenizing data. C error: Expected 2 fields in line 2, saw 4
We receive a ParserError that tells us pandas expected 2 fields (since this was the number of columns in the first row) but it saw 4.
This error tells us that the max number of columns in any given row is 4.
Thus, we can import the CSV file and supply a value of range(4) to the names argument:
import pandas as pd #import CSV file with differing number of columns per row df = pd.read_csv('uneven_data.csv', header=None, names=range(4))) #view DataFrame print(df) 0 1 2 3 0 A 22 NaN NaN 1 B 16 10.0 12.0 2 C 25 10.0 NaN 3 D 14 2.0 7.0 4 E 20 4.0 NaN
Notice that we’re able to successfully import the CSV file into a pandas DataFrame without any errors since we explicitly told pandas to expect 4 columns.
By default, pandas fills in any missing values in each row with NaN.
If you’d like the missing values to instead appear as zero, you can use the fillna() function as follows:
#fill NaN values with zeros df_new = df.fillna(0) #view new DataFrame print(df_new) 0 1 2 3 0 A 22 0.0 0.0 1 B 16 10.0 12.0 2 C 25 10.0 0.0 3 D 14 2.0 7.0 4 E 20 4.0 0.0
Each NaN value in the DataFrame has now been replaced with a zero.
Note: You can find the complete documentation for the pandas read_csv() function here.
The following tutorials explain how to perform other common tasks in Python: