ML Zoomkcamp Capstone Project 1

The Dataset for this work is from Zindi Africa Financial Inclusion in Africa Challenge

The objective of the competition is to create a machine learning model to predict which individuals are most likely to have or use a bank account.

You are asked to predict the likelihood of the person having a bank account or not (Yes = 1, No = 0).

Please note that all the code in this repository is written and tested on a Linux machine

OS: Ubuntu 20.04 LTS
Python: 3.10
pip: 22.2.1
Docker version: 20.10.17, build 100c701

Setup

Follow the setup instructions below

Build Docker Image

docker build --tag capstone-project .

Run image as a container

To locate our image with the tag you created above, run the command below

docker images

Choose the image you want to run and execute the docker run command followed by the image name

docker run -it --rm -p 8080:8080 capstone-project:latest

After succesfully runing the command above, you will see that docker is running.

Test

Open a new terminal window with the app by running in another window and run the command below:

python3 test.py
Finally, Upload the docker image to DockerHub or Amazon ECR using the docker push <image_name> command

Deployment on AWS Lambda

Go to AWS Console and Lambda
Create a function
Choose container image and fill in the necessary details
Select browse image to choose from our list of images
Finally, choose create function

The lambda function will now be created, we can go ahead to start testing it. Follow the instructions in this video to learn how to deploy a docker image to AWS Lambda

Deployment on Kubernetes

You can learn how to deploy kubernetes service using by following this ML Zoomcamp Video

Setup local Kubernetes cluster by installing kubectl & kind

create deployment

Deployment configuration can be found in the deployment.yaml file
kind load docker-image <image_name>
kubectl apply -f deployment.yaml
kubectl get deployment
kubectl port-forward <port_name>

Create Service

configuration can be found in the service.yaml file
kubectl apply -f service.yaml
kubectl get service
kubectl port forward service/<service_name> 8080:80

Files

Dataset: Train.csv, Test.csv, SampleSubmission.csv, VariableDefinition.csv
Model: zindi-fi.bin, dv.bin
Prediction: lambda_fucntion.py, test.py
Jupyter Notebook: Zindi_Financial_Inclusion.ipynb, predict-test.ipynb
Other: Dockerfile deployment.yaml service.yaml