Rinse_Over_Run

Rinse Over Run:

Forecasting turbidity in F&B industrial equipment.
https://www.drivendata.org/competitions/56/predict-cleaning-time-series/

Authors:

The BI Sharpes: Pat Walsh & David Belton
March 2019
Pat's Github

Generating submission:

1. Copy the contest data to /data directory. This data should be available at: https://www.drivendata.org/competitions/56/predict-cleaning-time-series/page/125/#datasets

2. Ensure the requirements in requirements.txt are satisfied.

   • I'm running this on Windows but this will work in python 3.7.1 with the requirements met.

3. Getting Predictions:

   • navigate to Rinse_Over_Run base folder
   • Run following python script: Rinse_Over_Run.py

4. The recreated final submission is stored in data/test_predictions.csv

Details about generating predictions:

When the script is running, you should see:

(base) C:\Users\YourNameHere\Documents\GitHub\Rinse_Over_Run>python Rinse_Over_Run.py
Begining data load this may take a few minutes
train_values.csv uploaded
recipe_metadata.csv uploaded
train_labels.csv uploaded
test_values.csv uploaded
Launching
Training Data prep successful
Test Data prep successful
Object Flow1 successful
Clustering Flow2 successful
Rinse Over Run successful: prediction can be found at data/test_predictions.csv
...It's been done.

Heres an explanation of these progress updates:

Begining data load this may take a few minutes.

• confirming required libraries have been imported and functions have been defined correctly

train_values.csv uploaded

• train_values.csv found and read

recipe_metadata.csv uploaded

• recipe_metadata.csv found and read

train_labels.csv uploaded

• train_labels.csv found and read

test_values.csv uploaded

• test_values.csv found and read

Launching

• other variables defined and the model workflow is being launched

Training Data prep successful

• Prepped the training data for the Flows

Test Data prep successful

• Prepped the test data for the Flows

Object Flow1 successful

• Object Flow1 executed succesfully

Clustering Flow2 successful

• Cluster Flow2 executed succesfully

Rinse Over Run successful: prediction can be found at data/test_predictions.csv
...It's been done.

• Verifying completion

...It's been done.
Referencing a classic Simpsons quote.

It takes about 10 minutes to for the script to execute and recreate the solution, the most time consuming part is uploading the csv's as they are quite large at 2.13GB.