# You can use "astype" method
# suppose you want to correct your "sales" column data type
df['sales'] = df['sales'].astype('float64')
>>> df.astype({'col1': 'int32'}).dtypes
col1 int32
col2 int64
dtype: object
df = pd.read_csv("weather.tsv", sep="\t",
dtype={'Day': str,'Wind':int64})
df.dtypes