Untitled

Just published an article on Using Bagging And XGBoost To Train Large Datasets. The link to the article is shown below.

Just completed the ML Zoomcamp 2024 Competition on kaggle which is a Retail Demand Forecasting Challenge! The competition challenges participants to forecast customer demand based on historical sales and related data from a retailer.

The dataset consists of 10 files in csv format with a total size of 872.28 MB and the files had to be aggregated to form a single dataset of 7,432,685 rows, and 16 columns.

Creating a model for this dataset was a challenge because the size was to large for all the platforms I used so, I have to apply an ensemble method where a fraction of the date was modelled in series and later combined at the end.

This was a wonderful experience that was much more different from the toy dataset that we are used to.

ML Zoomcamp Project Capstone 3 – 3: First trainings

After conducting the exploratory data analysis, 3 pre-trained deep convolutional neural network models were used to train the data for 10 epochs each with the aim of finding the one that will return the best test accuracy which will then be used to create the final model. The results obtained with BATCH_SIZE = 32, LEARNING_RATE = 0.0001 are shown below:

Xception - 98.93%

InceptionV3 - 98.32%

ResNet101V2 - 96.80%

ML Zoomcamp Project Capstone 2 – 2: Exploratory Data Analysis

The dataset used in this project is the Brain Tumor MRI Dataset https://www.kaggle.com/datasets/masoudnickparvar/brain-tumor-mri-dataset from kaggle.

A brain tumor is a mass of abnormal cells that grows in the brain, it can be benign (non-cancerous) or malignant (cancerous). Its symptoms are:

Headaches, which can be severe, persistent, or come and go

Seizures, which can be mild or severe

Weakness or paralysis in part of the body

Loss of balance

Changes in mood or personality

Changes in vision, hearing, smell, or taste

Nausea and vomiting

Difficulty speaking

Difficulty swallowing

The growth of brain tumors can cause the pressure inside the skull to increase leading to brain damage, and loss of life if not discovered early and properly treated.

Just completed the tenth week of Machine Learning Zoomcamp.

The lessons covered include:

TensorFlow Serving

Creating a pre-processing service

Running everything locally with Docker-compose

Introduction to Kubernetes

Deploying a simple service to Kubernetes

Deploying TensorFlow models to Kubernetes

Deploying to EKS

Just completed the ninth week of Machine Learning Zoomcamp.

The lessons covered include:

Introduction to Serverless

AWS Lambda

TensorFlow Lite

Preparing the code for Lambda

Preparing a Docker image

Creating the lambda function

API Gateway: exposing the lambda function

Just completed the eighth week of Machine Learning Zoomcamp.

The lessons covered include:

Fashion classification

Setting up the Environment on Saturn Cloud

TensorFlow and Keras

Pre-trained convolutional neural networks

Convolutional neural networks

Transfer learning

Adjusting the learning rate

Checkpointing

Adding more layers

Regularization and dropout

Data augmentation

Training a larger model

Using the model

Just completed the sixth week of Machine Learning Zoomcamp.

The lessons covered include:

Credit risk scoring project

Data cleaning and preparation

Decision trees

Decision tree learning algorithm

Decision trees parameter tuning

Ensemble learning and random forest

Gradient boosting and XGBoost

XGBoost parameter tuning

Selecting the best model

Just completed the fifth week of Machine Learning Zoomcamp.

The lessons covered include:

Saving and loading the model

Web services: introduction to Flask

Serving the churn model with Flask

Python virtual environment: Pipenv

Environment management: Docker

Deployment to the cloud: AWS Elastic Beanstalk

Just submitted my final project for LLM Zoomcamp 2024. It has been a wonderful experience because the course gave me the opportunity to work with both closed and open source LLMs. The knowledge attained include vector-search, monitoring, orchestration and best practices. My sincere gratitude goes to Alexey Grigorev for making all these possible.

The link to the project is below:

Just completed the forth week of Machine Learning Zoomcamp.

The lessons covered include: 4.1 Evaluation metrics: session overview 4.2 Accuracy and dummy model 4.3 Confusion table 4.4 Precision and Recall 4.5 ROC Curves 4.6 ROC AUC 4.7 Cross-Validation

Just completed the third week of Machine Learning Zoomcamp.

The lessons covered include: 3.1 Churn prediction project 3.2 Data preparation 3.3 Setting up the validation framework 3.4 EDA 3.5 Feature importance: Churn rate and risk ratio 3.6 Feature importance: Mutual information 3.7 Feature importance: Correlation 3.8 One-hot encoding 3.9 Logistic regression 3.10 Training logistic regression with Scikit-Learn 3.11 Model interpretation 3.12 Using the model 3.13 Summary

Just completed the second week of Machine Learning Zoomcamp.

The lessons covered include: 1 Car price prediction project 2 Data preparation 3 Exploratory data analysis 4 Setting up the validation framework 5 Linear regression 6 Linear regression: vector form 7 Training linear regression: Normal equation 8 Baseline model for car price prediction project 9 Root mean squared error 10 Using RMSE on validation data 11 Feature engineering 12 Categorical variables 13 Regularization 14 Tuning the model 15 Using the model 16 Car price prediction project summary 17 Explore more

Just completed the first week of Machine Learning Zoomcamp.

The lessons covered include:

What is Machine Learning

Machine Learning vs Rules Based Systems

Supervised Machine Learning

CRISP-DM

Model Selection Process

Setting up the Environment

Introduction to NumPy

Linear Algebra Refresher

Introduction to Pandas

Summary

Just completed the fifth week of LLM Zoomcamp.

The lessons covered include:

Review

Ingest

Chunk

Tokenization

Embed

Export

Retrieval

Trigger Daily Runs

Just completed the forth week of LLM Zoomcamp.

The lessons covered include:

Introduction to monitoring answer quality

Evaluation and Monitoring in LLMs

Offline RAG Evaluation

Offline RAG Evaluation: Cosine Similarity

Offline RAG Evaluation: LLM as a Judge

Capturing User Feedback

Monitoring the System