nimble-tk 1.0.2

A collection of simple but powerful utilities for Data Analytics and Data Science workflows.

nimble toolkit

A collection of simple but powerful utilities for ML and Data Analytics workflows.

Installation

pip install nimble-tk

Examples

Logging

The below set of cells give some utility wrappers on python's logging functionality. The following line will create a log file at /tmp/try_nimble.log

• As the logs get written, the log file will go up to max 50 MB in size by default and then will get rolled over.
• At most, 10 such files are maintained before removing the earliest file.
• Logs will also be written to the console (in this case the notebook console)

ntk.init_file_logger(log_file_path='/tmp/try_nimble.log', console_log_on=True)

Output:

2023-12-09 11:10:43.270595 :: MainThread ::  Log file init at /tmp/try_nimble.log

Example of an info log:

ntk.log_info("some log message")

Output:

2023-12-09 11:10:43.289434 :: MainThread ::  some log message

Example of an error log:
Catching an exception and logging the stack trace:

try:
    tmp = 1/0
except:
    ntk.log_traceback("Error while running operation")

Output:

2023-12-09 11:10:43.342127 :: MainThread ::  Error while running operation :: ZeroDivisionError: division by zero ::
 Traceback (most recent call last):
  File "/tmp/ipykernel_3709/4084614229.py", line 2, in <module>
    tmp = 1/0
ZeroDivisionError: division by zero

All logs are also written to the log file:

! tail -10 /tmp/try_nimble.log

Output:

2023-12-09 11:10:43,268 MainThread logger.py:88:  Log file init at /tmp/try_nimble.log
2023-12-09 11:10:43,284 MainThread logger.py:88:  some log message
2023-12-09 11:10:43,311 MainThread logger.py:88:  some error message
2023-12-09 11:10:43,334 MainThread logger.py:88:  Error while running operation :: ZeroDivisionError: division by zero ::
 Traceback (most recent call last):
  File "/tmp/ipykernel_3709/4084614229.py", line 2, in <module>
    tmp = 1/0
ZeroDivisionError: division by zero

Logs can also be directly written to the file without logging on the console. This helps in cases where we do not want to flood the console with a lot of logs.

ntk.log_info_file("some log message")
ntk.log_error_file("some error message")

try:
    tmp = 1/0
except:
    ntk.log_traceback_file("Error while running operation")

! tail -7 /tmp/try_nimble.log

Output:

2023-12-09 11:10:43,501 MainThread logger.py:88:  some log message
2023-12-09 11:10:43,505 MainThread logger.py:88:  some error message
2023-12-09 11:10:43,510 MainThread logger.py:88:   :: ZeroDivisionError: division by zero ::
 Traceback (most recent call last):
  File "/tmp/ipykernel_3709/3515687902.py", line 5, in <module>
    tmp = 1/0
ZeroDivisionError: division by zero

Concurrent Processing

The ntk.run_concurrently(...) method provides a wrapper around python's multi-processing and multi-threading functionality.

Set up a list of functions to execute:

def analytic_function(customer_id):
    ntk.log_info_file(f"Running analytic_function for customer_id: {customer_id}")
    
    if customer_id == 0:
        # raising an error here to demonstrate error handling
        raise ValueError(f"Invalid customer id {customer_id}")
        
    # - Query the DB
    # - Do feature engineering
    # - Run other analytics
    
    # - Write the result data to disk or return a DF
    return pd.DataFrame({
        'CUSTOMER_ID': [customer_id]*5,
        'OTHER_DATA': np.random.randint(0, 9, 5)
    })

functions = []
for customer_id in range(10):
    functions.append((analytic_function, {'customer_id': customer_id}))

Run the functions concurrently with the following lines of code:

results, errors = ntk.run_concurrently(functions, max_workers=ntk.get_num_cpus(), fork=True)

df_result = pd.concat([result[1] for result in results])
df_result.head(index=False)

fork=True will launch multiple processes and is recommended to be used for CPU-bound tasks.

fork=False will launch multiple threads, should to be used for IO-bound tasks.

Output:

2023-12-09 11:10:46.767042 :: MainThread ::  Ran: 10 functions - Successfull: 9, Failed: 1

CUSTOMER_ID	OTHER_DATA
1	3
1	0
1	3
1	1
1	3
...	...