So I’m learning how to use pandas and I’m trying to group a df based on a set of parameters in one of the columns and perform the mean on another column.

The original df is this:

> df.head()

        USAF    YR--MODAHRMN    TEMP    MAX MIN     Celsius
0   28450   201705010000    31.0    NaN NaN -1.0
1   28450   201705010020    30.0    NaN NaN -1.0
2   28450   201705010050    30.0    NaN NaN -1.0
3   28450   201705010100    31.0    NaN NaN -1.0
4   28450   201705010120    30.0    NaN NaN -1.0

df.types

USAF              int64
YR--MODAHRMN      int64
TEMP            float64
MAX             float64
MIN             float64
Celsius         float64

The objetive is to convert the column YR–MODAHRMN into a string cut the portion that is of no interest and use the resulting column to group the data in order to calculate the mean of Celsius based on that grouping.

df["YR--MODAHRMN"]=df["YR--MODAHRMN"].astype(str)
df["YR--MODA"]=df["YR--MODAHRMN"].str[0:8]
df.head()
df.dtypes

    USAF    YR--MODAHRMN    TEMP    MAX MIN Celsius YR--MODA
0   28450   201705010000    31.0    NaN NaN -1.0    20170501
1   28450   201705010020    30.0    NaN NaN -1.0    20170501
2   28450   201705010050    30.0    NaN NaN -1.0    20170501
3   28450   201705010100    31.0    NaN NaN -1.0    20170501
4   28450   201705010120    30.0    NaN NaN -1.0    20170501

df = df[["USAF","YR--MODA","Celsius"]]
df["MeanC"] = df["Celsius"].groupby(df["YR--MODA"]).mean(numeric_only=True)
df.dtypes
df.head()

USAF          int64
YR--MODA     object
Celsius     float64
MeanC       float64
dtype: object

    USAF    YR--MODA    Celsius MeanC
0   28450   20170501    -1.0    NaN
1   28450   20170501    -1.0    NaN
2   28450   20170501    -1.0    NaN
3   28450   20170501    -1.0    NaN
4   28450   20170501    -1.0    NaN

All of the results are NaN eventhough the inputed data is numeric I’v ran pd.isnull(df)and it returned false on the Celsius column. What am I doing wrong?