AutoML

Bots to Auto generate Notebook for Baseline DataScience Model

Introduction

The phases of Data Science modeling process is largely fixed.

Stage of Data Science Modeling process

1. Fetch Dataset
2. Data Explore
3. Data Cleanup and Feature Engineering
4. Data Modeling
5. Advance Modeling (voting/stacking/blending)
6. Final Prediction

Our bot auto generates notebook code for these phases for provided problem data configuration

Salient features

• Generates Model for both learning type Regression & Classification
• Performs basic Hyperparameter tuninng of all generated models
• Provides Model-Leaderboard for model evaluation
• Covers wide range of Models ( trains data on total 13 model that includes 5 base models, 2 bagging models, 3 boosting models, 2 voting and 1 stacking model).
• Automates all generic steps hence saves Time
• Guides in further modeling decisions

Project Modules

1. automl_bot_local_notebook : AutoGenerate jupyter notebook on local machine

Usage Instruction : Provide/update following configuration parameter inside automl_bot_local_notebook/runconfig.py and execute

• "_TRAIN_FILE" -> Name of training dataset file
• "_TEST_FILE" -> Name of test dataset file
• "_EXT" -> Dataset file type extension, currently supports csv and xlsx
• "_DATASET_FOLDER_PATH" -> Absolute path of folder where test and train files are present
• "_TARGET_COL" -> Name of target feature to be predicted
• "_ID_COL" -> Nmae of unique id column or column which will exist inside final submission file
• "_METRIC" -> evaluation metric name (as per sklearn library)
• "_TYPE" -> Learning type either regression or classification

2. automl_bot_kaggle_kernel : Autogenerates kaggle kernel

Usage Instruction :

Update your kaggle id under id (in line 17) of kaggle_bot.py and provide/update following configuration parameter inside automl_bot_kaggle_kernel/runconfig.py and execute

• "_COMPETITION" > Kaggle competition name
• "_TARGET_COL" -> Feature to be predicted
• "_ID_COL" -> Unique id column or column which will exist inside final submission file
• "_NAME" -> Name of kaggle kernel the bot is generating
• "_METRIC" -> evaluation metric
• "_TYPE" -> Learning type either regression or classification
• "_TEST_FILE" -> Test File name ( if other than test)
• "_TRAIN_FILE" -> Training file name ( if other than train)

3. automl_bot_google_colab : Autogenerates google colab notebook

Usage instruction : tobedone

Future work

• Automate Deep learning models e.g. image processing
• NLP i.e. support for text features
• Optimization

References :