azureml_dataprep_csv_load_column_filter_to_pandas.py

1import azureml.dataprep as dprep
2
3# Load data from a CSV file
4dflow = dprep.read_csv(path='https://dprepdata.blob.core.windows.net/demo/Crime_Data_from_2010_to_Present.csv')
5
6# Select specific columns
7dflow = dflow.keep_columns(['DR_NO', 'Date Reported', 'Victim Age', 'Victim Sex'])
8
9# Filter rows where Victim Age is greater than 0
10dflow = dflow.filter(dflow['Victim Age'] > 0)
11
12# Pull the data into a local Pandas DataFrame
13df = dflow.to_pandas_dataframe()
14
15# Display the first 5 rows
16print(df.head())