setup complete. most of the code will be uploaded at riyadhrazzaq/pywork or at kaggle/riyadhrazzaq
Read up on Decision Tree Regressor theory. https://www.saedsayad.com/decision_tree_reg.htm
Stuck in implementation of Decision Tree Regressor. will updated if completed.
Decision Tree Regressor Implementation. SCRIPT THEORY
Decision Tree Classifier implementation. [SCRIPT] [theory]
SVM Theory [theory]
Quadratic Optimization for SVM, resources-
- https://courses.csail.mit.edu/6.867/wiki/images/a/a7/Qp-cvxopt.pdf
- https://scaron.info/blog/quadratic-programming-in-python.html
SVM Implementation(binary, linear kernel) [theory] [SCRIPT]
Quadratic Programming code/format taken from https://pythonprogramming.net/soft-margin-kernel-cvxopt-svm-machine-learning-tutorial/ , they have credited http://mblondel.org/ for the original code.
Netflix Dataset EDA
Netflix EDA Complete
Netflix Recommender. Cosine Similarity with Word Embedding, TfIdf.
Netflix titles analysis and recommender based on clustering. [NOTEBOOK]
Collaborative Filtering with SGD. [THEORY]
Spent the whole day trying to fix CF w/ SGD implementation from scratch. I made a mistake in SGD. Will fix later in sha Allah.
As I was stuck because of simple SGD implementation, I realized I need backward propagation for a minute. Here is Linear Regression
w/ Gradient Descent [NOTEBOOK]
w/ Stochastic Gradient Descent [SCRIPT]
Non-negative Matrix Factorization w/ SGD. [NOTEBOOK]
Ridge Regression. [SCRIPT]
Lasso Regression [NOTEBOOK] This implementation is really buggy. But I did not understand the theory clearly to code further. Maybe later, in sha Allah.
Gaussian Naive Bayes [NOTEBOOK]. Really proud of this implementation. Spent the whole day on Bayes Theorem, Normal Distribution. 3Blue1Brown's video on Bayes Theorem helped the most. Never seen such an intuitive video about probability. This is the first code in ML that I have written had the lowest number of bugs. 1, actually. Forgot to add tmp+=1
inside function def predict()
in class GaussNB
Multinomial Naive Bayes implementation. Can't fix a bug.
This gave a lot of trouble. Lesson learned- read the theory from multiple sources if possible, even after you understand it. Multinomial NB implementation complete. [NOTEBOOk] with Spam Classification. [SCRIPT]
Logistic Regression theory from ISLR and ESL. Plan to read this too.
Logistic Regression from Scratch using Batch Gradient Descent. [NOTEBOOK]
Linear Algebra [MITOPENCOURSEWARE]. EDA. will post link if I work on it further.
ICU Prediction of COVID-19 Patients. [NOTEBOOK]
Linear Regression, Python for Data Analysis exercises on Pandas.
Cross Validation. Notebook on DrivenData competition DengAI
Bagging, Random Forest, DengAI, Boosting(AdaBoost)
Best Subset Selection notebook partially done. Practiced few sections from Python for Data Analysis, Wes McKinney.
Best Subset complete. [NOTEBOOK]
Hyperparameter tuning with RandomSearchCV from scratch. [NOTEBOOK]
Handling Multiclass Classification with 1vsRest, 1vs1 from scratch. [NOTEBOOK]
House Price Revisited.
Hierarchical Clustering. Theory from ISLR book. Code partially done.
Hierarchical Clustering - Agglomerative approach, code complete for linkage types: single, complete.
Feature Preprocessing and feature generation review. Currently reading chapter 4 of ILA by Prof. G. Strang.