OPEN SOURCE PROJECTS
IVY - let's unify ai
PROJECT DESCRIPTION:
Ivy is a Machine Learning framework, with a mission to unify all ML frameworks and allow code transpilations between them. It currently supports four different ML frameworks, Tensorflow, Numpy, JAX, and MXNET.
​
CONTRIBUTION:
I Implemented a function for Tensorflow 'Frontend API'. (Frontend APIs are the supported frameworks that Ivy can convert code from).
​
It was a Linear Algorithm function (Cholesky) used to decompose a tensor. These 'Frontend APIs' play an important role during code transpilation.
​
PULL REQUEST: https://github.com/unifyai/ivy/pull/9115
​
​
TEAM DISCUSSION:
Contribution and team discussion were all remote. Ivy has a Large Discord community that is there to help in understanding the project and giving guidance on the contribution process.
​
PROJECTS
CALIFORNIA STATE HOUSE PRICE PREDICTION
PROJECT DESCRIPTION:
This project will be solving a Supervised Machine Learning Problem. Real Estate Companies rely on data analysis and prediction to make smart investments. To help such a company achieve its goal, I am going to Implement a Random Forest Machine Learning algorithm on a previous house census dataset to build a predictive model. The model is going to predict the house value depending on a number of independent features. From such a model, the company will be able to predict house values in one location and compare them with the house values in other locations in the market. This will prompt the decision of where to make investments, withdraw, or hold on for some time.
​
DEMONSTRATION: TRY IT!!!
Link the to the Web Application: https://house-value-predicto.herokuapp.com/
​
HOW TO USE THE APPLICATION:
Open the Application link
​
Input Prediction entries (to find relevant entries go to my GitHub repo https://github.com/14Emanuel/California-house-price-prediction and open the file 'testing phot dataset.png')
​
Press the Predict button
​
The Output button will show you the predicted house value
​
TECHNOLOGIES USED:
Data Science: Python, Supervised Machine Learning Algorithms, Scikit Learning, Pandas and Numpy, Matplotlib and Seaborn, Jupyter notebooks
​
Website UI: HTML, CSS
​
Website Backend: Python Flask Framework
​
Cloud Application Platform: Heroku Cloud Platform
ARTICLE SUMMARIZER
PROJECT DESCRIPTION:
This project apparently will be solving a personal problem. I normally stay updated about global news by reading articles and blogs from new outlets e.g BBC news, DW news, and new york times. The articles are verbose and make reading as many articles as possible slower. I decided to use my Natural Language Processing skills to build an application that will summarize any article or blog in only 4 sentences. This has been an Instant success to my morning schedule since now I can read an average of 20 articles in 30 minutes instead of 10. This application will help me increase my global awareness through articles even more.
​
DEMONSTRATION: TRY IT!!!
Link to the Web Application: https://text-summarizer.up.railway.app/
​
HOW TO USE THE APPLICATION:
Open the Application link
​
Type or paste your article in the text area shown in the App
Press the Summarize button
​
Output: You are going to be redirected to an output page where you will find your summarized article
​
TECHNOLOGIES USED:
Data Science: Python, Jupyter, Spacy
​
Website UI: HTML, CSS
​
Website Backend: Python Flask Framework
​
Host: Docker images, Railway Cloud
WHOLESALE CUSTOMER SEGMENTAION
PROJECT DESCRIPTION:
This project will be solving an unsupervised Machine Learning Problem. Wholesalers deals with a large variety of products and a high number of customers. In order for their marketing team to run targeted customer advertisements; they need the help of a Data Scientist. I am going to solve their problem by building a clustering model. The model is going to cluster customers and segment them according to the products they purchase on a regular basis. With the application in place, the marketing team will now be able to recommend the right product to its customers during an advertisement campaign. This practice will enhance customer retention, and brand identity, and improve the Channel of Distribution.
DEMONSTRATION TRY IT!!!:
Link to the Web App: https://customer-segmentation-production.up.railway.app/
​
HOW TO USE THE APPLICATION:
Open the Application link
​
Input Prediction entries (to find relevant entries go to my GitHub repo: https://github.com/14Emanuel/Wholesale-Customer-Segmentation and open the file 'test photo dataset.png')
​
Press the Predict button
​
The Output button will show you the predicted Customer Segment
​
TECHNOLOGIES USED:
Data Science: Python, KMeans, Pandas and Numpy, Matplotlib and Seaborn, Jupyter notebooks
​
Website UI: HTML, CSS
​
Website Backend: Python Flask Framework
​
Host: Docker images, Railway