CODEWITHSUNDEEP
pandasnumeric
user2100039
user2100039

Reputation: 1356

Pandas Convert Single Column to Numeric

apparently, there is a column in my pandas DF - ncData - that is not numeric because calcuations that I'm tryting to do give me an error like "no numeric data to aggregate". So, I have seen the documentation https://appdividend.com/2020/06/18/pandas-to_numeric-function-in-python-example/ and have tried some things but nothing is working yet. My data looks like this and it is a panda DF--

    plant_name business_name maint_region_name wind_speed_ms dataset  year  month  day  hour
0  RIO DO FOGO        BRAZIL            BRAZIL          7.88    ERA5  2021      5   31    21
1  RIO DO FOGO        BRAZIL            BRAZIL          7.95    ERA5  2021      5   31    20
2  RIO DO FOGO        BRAZIL            BRAZIL          7.72    ERA5  2021      5   31    19

The column "wind_speed_ms" apparently is not numeric. How do i change this column to numeric? I'v tried this but how do i assign the results or the conversion back to the original df?

pd.to_numeric(ncData.wind_speed_ms, errors='raise', downcast=None)

should i asssign it a new name for the output and then re-insert it back into "ncData"? thank you,

Upvotes: 1

Views: 354

Answers (1)

Aaron Diamond-Reivich
Aaron Diamond-Reivich

Reputation: 643

One common reason for pandas not recognizing columns as numbers when you expect it to be a number is the result of number formatting. For example, having commas in a series of numbers might cause pandas to recognize the series as an object instead of a number.

If your data set isn't toooo large, one way to look for characters that might be causing your series to be registered as a object is to use

df['wind_speed_ms'].unique().tolist()

and then scanning the list for commas, parens, etc.

Once you identify the character that is causing issues, you can replace it for a valid character or remove it all together.

You could so something like this:

df['wind_speed_ms'] = df['wind_speed_ms'].apply(lambda x: x.replace(',', '.'))

Also, btw: I am the creator of a python packaged called mitosheet, a spreadsheet extension to Jupyter that converts all of your edits into valid python code. I actually just implemented a feature that lets you convert the type of your columns with the click of a button.

Upvotes: 1

Related Questions