This repository contains a collection of Google Colab notebooks demonstrating various machine learning and data mining tasks using the PyCaret low-code machine learning library, fulfilling all requirements of the assignment.
The primary goal of this assignment is to demonstrate proficiency in using PyCaret's AutoML capabilities to solve six distinct data mining tasks. All code is original (rewritten) and executes successfully in the provided Colab notebooks, using dedicated datasets unique to this submission.
- PyCaret Version: All notebooks use
pycaret==2.3.5. - Code Integrity: All code has been rewritten (not copied from official PyCaret examples) to ensure originality.
- Dataset Selection: Each task uses a unique dataset sourced from Kaggle or other public repositories, distinct from those used in official PyCaret tutorials.
- AutoML Utilization: Full use of PyCaret's AutoML features, including the
compare_models()andtune_model()functions, is demonstrated. - GPU Acceleration: Where supported for Classification and Regression tasks, GPU acceleration is enabled using
setup(use_gpu=True)to leverage Google Colab's GPU runtime.
The repository is organized to clearly present each task and its corresponding Colab notebook.
| Task | PyCaret Module | Notebook Filename | Dataset Focus | Execution Notes |
|---|---|---|---|---|
| Binary Classification | pycaret.classification |
1_Binary_Classification.ipynb |
Predict two target classes. | GPU enabled (use_gpu=True). |
| Multiclass Classification | pycaret.classification |
2_Multiclass_Classification.ipynb |
Predict three or more target classes. | GPU enabled (use_gpu=True). |
| Regression | pycaret.regression |
3_Regression.ipynb |
Predict a continuous numerical value. | GPU enabled (use_gpu=True). |
| Clustering | pycaret.clustering |
4_Clustering.ipynb |
Unsupervised grouping of data points. | Successfully executed. |
| Anomaly Detection | pycaret.anomaly |
5_Anomaly_Detection.ipynb |
Identify rare items/events/observations. | Successfully executed. |
| Time Series Forecasting | pycaret.time_series |
7a_TS_Univariate.ipynb |
Forecasting a single variable without external data. | Successfully executed. |
- All notebooks are designed to be cloned and executed successfully in the target environment (Google Colab).
- The output cells in each notebook show successful execution, demonstrating the use of
setup(),compare_models(), and model deployment functions. - GPU usage is confirmed in the output of the
setup()function where applicable.
A long-form video tutorial has been provided, which serves as a detailed walkthrough for every Colab notebook in this repository.
- Walkthrough Requirement: The video walks through each of the 6 Colab notebooks sequentially.
- Explanation: Each notebook's purpose, dataset, key PyCaret functions (
setup,compare_models,create_model,tune_model), and final output are explained in approximately one minute. - Proof of Execution: The video explicitly shows the code running in the student's Google Colab environment, confirming successful execution and results for all tasks.
Video Link: https://drive.google.com/file/d/1TbgP-W6n31U8YXIPhgfICuaT7g9Bq6Pk/view?usp=drive_link https://drive.google.com/file/d/1Nt-iubejLP0utEmYTD0QvHTdl0AedaNq/view?usp=drive_link https://drive.google.com/file/d/1oLqUYGuiiM2QQ3mWdLg5IOzgvJtNFJ3a/view?usp=drive_link https://drive.google.com/file/d/1OaoWqpaYIzrGovBk1gcvB75lv_wBsrvO/view?usp=drive_link https://drive.google.com/file/d/1Sgc-fYVjib57R1JKjUVXRL_NO97I5PaT/view?usp=drive_link https://drive.google.com/file/d/1Ih1p2bqqGgH_9GXjlPlwIZNsb7uaigcS/view?usp=drive_link