Morris Lee
k-w-lee
I love programming. An explorer of curiosity
Languages
Loading contributions...
Top Repositories
In this project, we have used linear programming to develop a diet optimization system to satisfy constraints of nutrition goal with minimum ingredient cost
Deployed machine learning models & data science calculators at morris-lee.com using Python & Heroku.
View the deployed portfolio this link: https://k-w-lee.github.io/morris_lee.github.io/index.htmlMy Portfolio
Built a convolutional neural network, including recent variations such as residual networks; apply convolutional networks to visual detection and recognition tasks; and use neural style transfer to generate art and apply these algorithms to a variety of image, video, and other 2D or 3D data
Morris Lee Portfolio
🦜🔗 Build context-aware reasoning applications
Repositories
46Morris Lee Portfolio
🦜🔗 Build context-aware reasoning applications
Framework for orchestrating role-playing, autonomous AI agents. By fostering collaborative intelligence, CrewAI empowers agents to work together seamlessly, tackling complex tasks.
A simple screen parsing tool towards pure vision based GUI agent
Make websites accessible for AI agents
View the deployed portfolio this link: https://k-w-lee.github.io/morris_lee.github.io/index.htmlMy Portfolio
This repository contains code for training and using a text generation model based on the MinGPT library by Andrej Karpathy.
This is a classification machine learning task (APS Failure at Scania Trucks) that involved 171 attributes and 60000 observations.
In this project, we have used linear programming to develop a diet optimization system to satisfy constraints of nutrition goal with minimum ingredient cost
Deployed machine learning models & data science calculators at morris-lee.com using Python & Heroku.
Generates vCards based on a provided DataFrame and dictionary of column mappings. The script includes various functions to preprocess the data, validate the column mappings, clean and format the contact information, add a prefix to first names, and create and write the vCard entries to a file.
No description provided.
This notebook shows in-depth look on classifying images of children and adult using Deep Convolutional Neural Network. Given a limited size of images, it is not enough for model to learn to predict accurately on unseen data. The solutions are using Image Augmentation and Transfer Learning to improve the predictive power. Asides from that, this notebook also show approach to tune the learning rate, which is the most important hyper-parameter to be calibrated. As a result, the classifying performance achieve over 0.8 accuracy. Besides, this notebook also visualise what computer see in the convolutional images, and as well allow YOU to test your own images with the calibrated model.
This notebook will show 6 kinds of visualisation idioms, namely Donut Charts, Hat Graphs, Bar of Pie, Histogram, Violin and Box Plot. Hope this notebook can help you to uncover interesting findings of the math score raw data.
In this notebook, n-gram will be used for feature extraction, which is the first NLP approach that introduced by Markov in 1913. An N-gram is an N-character slice of a longer string. The intuition of the n-gram model is that instead of computing a prediction based on entire corpus, one can approximate the prediction by only contiguous slices sequence of n words. To explain feature extraction using n-gram with a demonstration of the sentence, “The student is alone happily”. The number of n-gram features can be calculated by k-n+1, where k is the number of words. The result is a bag-of-n-grams model. Table below shows the demonstration of different n-gram feature representations.
No description provided.
This notebook will show interesting visualisations about population trends for countries and as well clusters. As a result, Top 15 Fastest, Slowest growing population countries visualisation idioms were plotted in order. Plus, due to COVID-19 happens in 2019, I was curious how is the population trend looks like. It seems to me that the number 1 is surprising. Let's discover it.
One may wonders what kind of pre-processing technique should apply that can generate best result? Such as: * ordinal encoding or one-hot-encode * standardisation or normalisation Meanwhile taking care of model parameters is adding another layer of difficulty. This notebooks has show case how a nested grid search work to tune the preprocessing technique and as well models'parameters. It combines grid search and Halving grid search technique for optimisation. After evaluating 3 regressors, Random Forest Regressor is the best model due to the best MAR, MSE, RMSE, R2 and Adj-R2. Notebook Author: Morris Lee <br> Date: 27-8-2022
No description provided.
No description provided.
No description provided.
No description provided.
Apple Stock Prices (2013-2018)
No description provided.
No description provided.
No description provided.
A tool for predictive analytic
No description provided.
Built a convolutional neural network, including recent variations such as residual networks; apply convolutional networks to visual detection and recognition tasks; and use neural style transfer to generate art and apply these algorithms to a variety of image, video, and other 2D or 3D data
Random Forest Classifier is trained on each of the n-gram feature representations. The best score achieved is 0.93 macro F1 on testing set using tuned Random Forest and unigram feature vectors