🔥 30-Day Machine Learning & Deep Learning Challenge

Welcome to my 30-Day Machine Learning & Deep Learning Challenge repository! This repository is a structured learning journey designed to take you from foundational concepts to advanced deep learning architectures in just 30 days. The goal is not only to understand the theory but also to implement every concept practically in Python, and be able to explain it clearly—even to a beginner.

---

🎯 Purpose

The purpose of this repository is to provide a complete, hands-on guide to Machine Learning (ML) and Deep Learning (DL) for learners of all levels. By following this challenge, you will:

• Understand the fundamental mathematics behind ML/DL concepts.
• Learn how to implement core algorithms from scratch in Python.
• Build a strong intuition for concepts through analogies and examples.
• Practice explaining topics in your own words to ensure deep comprehension.
• Have a reference that can be revisited, modified, and shared with others for educational purposes.

This repository is designed with volunteering and learning in mind, so anyone can follow along, experiment, and improve their ML/DL skills.

---

📚 Repository Structure

The repository is organized into 30 folders, one for each day of the challenge. Each folder contains:

• day_<number>_<topic>
  Example: day_01_loss_function

Inside each folder, you will find:

1. Python script (.py)
   • Contains full working examples of the topic.
   • Includes step-by-step implementation, code comments, and outputs.
2. ReadMe.md (optional for each day)
   • Explains the topic in plain English.
   • Includes analogies, mathematical explanations, and mini-exercises.
   • Designed so you could explain the concept to a 10-year-old after studying.

---

🗂 Example Folder Structure

ml_fundamentals_challenge/
├── day_01_loss_function/
│   ├── loss_function.py
│   └── README.md
├── day_02_gradient_descent/
│   ├── gradient_descent.py
│   └── README.md
├── day_03_regularization/
│   ├── regularization.py
│   └── README.md
...
├── day_30_final_project/
│   ├── final_project.py
│   └── README.md
└── README.md

---

🔑 Learning Approach

Each day follows a clear and structured path:

1. Theory
   • Full explanation of the concept with formulas.
   • Analogies to make abstract ideas intuitive.
2. Practical Implementation
   • Python code with detailed comments.
   • Example outputs to see results in action.
3. Verification & Reflection
   • Mini exercises to reinforce learning.
   • Encouragement to explain the topic to others to solidify understanding.

---

📈 30-Day Learning Roadmap

Week	Days	Focus Area	Key Topics
Week 1	Days 1-7	Deep Understanding of Loss & Gradient Descent	Loss Functions, Gradient Descent, Learning Rate, Momentum, Regularization
Week 2	Days 8-14	ML Foundations	Linear/Logistic Regression, Metrics, Decision Trees, Ensembles, Feature Engineering
Week 3	Days 15-21	Deep Learning Fundamentals	Perceptron, Neural Networks, Forward/Backward Propagation, Optimization
Week 4	Days 22-30	Advanced Deep Learning	CNNs, RNNs, LSTM, Transformers, Final Project

Detailed Day-by-Day Plan

Week 1: Deep Understanding of Loss & Gradient Descent

Goal: Understand how models learn from the inside — loss, gradients, optimization steps.

Day	Topic	Theory	Practice	Goal	1-Minute LinkedIn Video
Day 1	Loss Function Mathematics	MSE, Cross-Entropy, formulas, meaning	Implement MSE and Cross-Entropy with numpy	Explain loss functions to a 10-year-old	https://www.linkedin.com/posts/serhii-kravchenko1_ai-ml-machinelearning-activity-7362854560731734016-pKAE/
Day 2	Introduction to Gradient Descent	Derivative as direction of smallest change	Implement gradient descent for one variable	Understand "rolling down the hill" analogy	https://www.linkedin.com/posts/serhii-kravchenko1_ai-ml-machinelearning-activity-7367575410915565568-1euG/
Day 3	Multidimensional Gradient Descent	Gradients for vectors and matrices	Implement gradient descent for linear regression	Calculate gradient step manually	https://www.linkedin.com/posts/serhii-kravchenko1_ai-ml-dl-activity-7368267543695740930-X2Wh/
Day 4	Learning Rate, Momentum, RMSProp	Why step size regulation matters	Add momentum to gradient descent	Master optimization techniques	https://www.linkedin.com/posts/serhii-kravchenko1_ai-ml-machinelearning-activity-7368644528934797313-AKas/
Day 5	Regularization	L1, L2, Elastic Net, Dropout	Add L2 regularization to linear regression	Understand "penalty for complexity"	https://www.linkedin.com/posts/serhii-kravchenko1_ai-machinelearning-deeplearning-activity-7369021320895963137-Tdqd/
Day 6	Practice: Gradient Descent + Regularization	Combine concepts	Build model on synthetic data	Experiment with hyperparameters	https://www.linkedin.com/posts/serhii-kravchenko1_ai-ml-artificialintelligence-activity-7369390120652808194-B7w8/
Day 7	Week 1 Explanation	Review and solidify	Explain all concepts in your own words	Deep comprehension check	https://www.linkedin.com/posts/serhii-kravchenko1_ai-artificialintelligence-machinelearning-activity-7369739519358672896-Ac2G/

Week 2: ML Foundations

Goal: Build foundation for classical algorithms.

Day	Topic	Theory	Practice	Goal	1-Minute LinkedIn Video
Day 8	Linear Regression	Formulas, MSE, gradient descent vs normal equation	Linear regression on Boston dataset	Master linear relationships	https://www.linkedin.com/posts/serhii-kravchenko1_ai-machinelearning-deeplearning-activity-7370093692860395520-1TJZ/
Day 9	Logistic Regression	Sigmoid, cross-entropy, gradient descent	Implement from scratch in Python	Understand classification basics	https://www.linkedin.com/posts/serhii-kravchenko1_ai-machinelearning-deeplearning-activity-7371156573056028672-r71Y/
Day 10	Classification Metrics	Accuracy, Precision, Recall, F1-score, ROC-AUC	Apply sklearn on simple classification	Evaluate model performance	https://www.linkedin.com/posts/serhii-kravchenko1_happy-day-10-of-our-ml-dl-challenge-activity-7371543490503053312-w63y/
Day 11	Decision Trees	Space partitioning, entropy, Gini	Build decision tree with sklearn	Understand tree-based decisions	https://www.linkedin.com/posts/serhii-kravchenko1_day-11-of-our-ml-challenge-decision-trees-activity-7373332438728589312-Xn8m/
Day 12	Ensembles: Random Forest, Gradient Boosting	Bagging vs Boosting concepts	Apply Random Forest on dataset	Master ensemble methods	https://www.linkedin.com/posts/serhii-kravchenko1_day-12-of-our-mldl-challenge-mastering-activity-7374071778450788352-5B7j/
Day 13	Feature Engineering & Scaling	Normalization, standardization, one-hot encoding	Preprocess real dataset	Prepare data for models	https://www.linkedin.com/posts/serhii-kravchenko1_day-13-of-our-ml-challenge-feature-engineering-activity-7374784863599599616-V38G/
Day 14	Week 2 Explanation	Review ML algorithms and preprocessing	Explain all concepts in your own words	Solidify ML fundamentals	https://www.linkedin.com/posts/serhii-kravchenko1_day-14-of-our-mldl-challenge-week-2-review-activity-7375539431258370048-qe8a/

Week 3: Deep Learning Fundamentals

Goal: Understand neural network structure and backpropagation.

Day	Topic	Theory	Practice	Goal	1-Minute LinkedIn Video
Day 15	Perceptron	Formula, linear combination, activation	Implement single perceptron	Understand "decision with weighted brain"	https://www.linkedin.com/posts/serhii-kravchenko1_day-15-of-our-mldl-challenge-perceptron-activity-7376610160917553152-L33t/
Day 16	Activation Functions	Sigmoid, ReLU, Tanh, LeakyReLU	Visualize functions on graphs	Master non-linearity	https://www.linkedin.com/posts/serhii-kravchenko1_day-16-of-our-deep-learning-challenge-activity-7378428512896065536-AwkL/
Day 17	Forward Propagation	Signal flow from input to output	Implement forward propagation with numpy	Understand information flow	https://www.linkedin.com/posts/serhii-kravchenko1_day-17-of-the-30-day-mldl-challenge-activity-7379484342286458880--qzJ/
Day 18	Backpropagation	Chain rule, gradients for each layer	Calculate gradients manually for one layer	Master "backward wave for corrections"	https://www.linkedin.com/posts/serhii-kravchenko1_%F0%9D%90%83%F0%9D%90%9A%F0%9D%90%B2-%F0%9D%9F%8F%F0%9D%9F%96-%F0%9D%90%81%F0%9D%90%9A%F0%9D%90%9C%F0%9D%90%A4%F0%9D%90%A9%F0%9D%90%AB%F0%9D%90%A8%F0%9D%90%A9%F0%9D%90%9A%F0%9D%90%A0%F0%9D%90%9A%F0%9D%90%AD-activity-7383473254843297792-zN_l/
Day 19	Optimization	SGD, Momentum, Adam	Implement Adam for simple NN	Advanced optimization techniques	https://www.linkedin.com/posts/serhii-kravchenko1_day-19-of-our-mldl-challenge-%F0%9D%90%8D%F0%9D%90%9E%F0%9D%90%AE%F0%9D%90%AB%F0%9D%90%9A%F0%9D%90%A5-activity-7386012096595066880-zyyT/
Day 20	Practice: NN on MNIST	Combine all concepts	Build 1-2 layer NN with numpy	End-to-end neural network	https://www.linkedin.com/posts/serhii-kravchenko1_%F0%9D%90%83%F0%9D%90%9A%F0%9D%90%B2-20-%F0%9D%90%A8%F0%9D%90%9F-%F0%9D%90%A8%F0%9D%90%AE%F0%9D%90%AB-%F0%9D%90%8C%F0%9D%90%8B%F0%9D%90%83%F0%9D%90%8B-%F0%9D%90%82%F0%9D%90%A1%F0%9D%90%9A-activity-7387467264029265920-yAJZ/
Day 21	Week 3 Explanation	Review NN, forward/backward propagation	Explain all concepts in your own words	Deep learning comprehension	https://www.linkedin.com/posts/serhii-kravchenko1_%F0%9D%90%83%F0%9D%90%9A%F0%9D%90%B2-21-%F0%9D%90%96%F0%9D%90%9E%F0%9D%90%9E%F0%9D%90%A4-3-%F0%9D%90%92%F0%9D%90%AE%F0%9D%90%A6%F0%9D%90%A6%F0%9D%90%9A%F0%9D%90%AB%F0%9D%90%B2-%F0%9D%90%A8%F0%9D%90%9F-activity-7388570578460364800-YxPp/

Week 4: Advanced DL & Modern Architectures

Goal: Understanding CNN, RNN, LSTM, Transformer, Attention.

Day	Topic	Theory	Practice	Goal	1-Minute LinkedIn Video
Day 22	CNN Basics	Convolution, Pooling, Flatten	Simple CNN on MNIST with PyTorch/Keras	Understand spatial processing	https://www.linkedin.com/posts/serhii-kravchenko1_%F0%9D%90%83%F0%9D%90%9A%F0%9D%90%B2-22-%F0%9D%90%A8%F0%9D%90%9F-%F0%9D%90%A8%F0%9D%90%AE%F0%9D%90%AB-%F0%9D%90%83%F0%9D%90%8B%F0%9D%90%8C%F0%9D%90%8B-%F0%9D%90%82%F0%9D%90%A1%F0%9D%90%9A-activity-7390046464795906050-PK25/
Day 23	Advanced CNN	Padding, stride, filter size	Visualize feature maps	Master convolutional operations	https://www.linkedin.com/posts/serhii-kravchenko1_%F0%9D%90%83%F0%9D%90%9A%F0%9D%90%B2-23-%F0%9D%90%A8%F0%9D%90%9F-%F0%9D%90%AD%F0%9D%90%A1%F0%9D%90%9E-30-%F0%9D%90%83%F0%9D%90%9A%F0%9D%90%B2-%F0%9D%90%83%F0%9D%90%8B%F0%9D%90%8C%F0%9D%90%8B-activity-7392570159795941376-lNaV/
Day 24	RNN Basics	Sequential data, hidden state	Simple RNN on synthetic sequence	Process sequential information	https://www.linkedin.com/posts/serhii-kravchenko1_%F0%9D%90%83%F0%9D%90%9A%F0%9D%90%B2-24-%F0%9D%90%A8%F0%9D%90%9F-%F0%9D%90%A8%F0%9D%90%AE%F0%9D%90%AB-%F0%9D%90%8C%F0%9D%90%8B%F0%9D%90%83%F0%9D%90%8B-%F0%9D%90%82%F0%9D%90%A1%F0%9D%90%9A-activity-7394390996068806656-hIT1/
Day 25	LSTM/GRU	Gates, memory cell	Implement LSTM with PyTorch/Keras	Handle long-term dependencies	https://www.linkedin.com/posts/serhii-kravchenko1_%F0%9D%90%83%F0%9D%90%9A%F0%9D%90%B2-25-%F0%9D%90%A8%F0%9D%90%9F-%F0%9D%90%A8%F0%9D%90%AE%F0%9D%90%AB-%F0%9D%90%83%F0%9D%90%8B%F0%9D%90%8C%F0%9D%90%8B-%F0%9D%90%82%F0%9D%90%A1%F0%9D%90%9A-activity-7396231676575662080-m949/
Day 26	Attention & Transformer	Self-attention, query/key/value	Simple transformer scheme	"Each word looks at others for context"	https://www.linkedin.com/posts/serhii-kravchenko1_%F0%9D%90%83%F0%9D%90%9A%F0%9D%90%B2-26-%F0%9D%90%A8%F0%9D%90%9F-%F0%9D%90%A8%F0%9D%90%AE%F0%9D%90%AB-%F0%9D%90%8C%F0%9D%90%8B%F0%9D%90%83%F0%9D%90%8B-%F0%9D%90%82%F0%9D%90%A1%F0%9D%90%9A-activity-7399468266399621120-UAH3/
Day 27	Optimization Tricks	BatchNorm, Dropout, LR scheduler, Gradient clipping	Add to CNN/RNN	Master training techniques	https://www.linkedin.com/posts/serhii-kravchenko1_%F0%9D%90%83%F0%9D%90%9A%F0%9D%90%B2-27-%F0%9D%90%A8%F0%9D%90%9F-%F0%9D%90%A8%F0%9D%90%AE%F0%9D%90%AB-30-%F0%9D%90%9D%F0%9D%90%9A%F0%9D%90%B2-%F0%9D%90%8C%F0%9D%90%8B%F0%9D%90%83%F0%9D%90%8B-activity-7402384778739478528-wtW1/
Day 28	Week 4 Summary	Summary of CNN, Advanced CNN, RNN, LSTM/GRU, Attention, Transformer, Optimization Tricks	Write a structured recap of all concepts learned during Week 4	Solidify understanding of all modern DL architectures	https://www.linkedin.com/posts/serhii-kravchenko1_github-serhii2009mlfundamentalschallenge-activity-7404915088186585088-Pjj0/
Day 29	Project: Emotion Detection	CNN for facial emotion recognition; data preprocessing; classification pipeline	Build an Emotion Detection CNN using FER2013 (or similar dataset): train, evaluate, visualize metrics	Apply deep learning to a real-world vision task	https://www.linkedin.com/posts/serhii-kravchenko1_%F0%9D%90%83%F0%9D%90%9A%F0%9D%90%B2-29-%F0%9D%90%A8%F0%9D%90%9F-%F0%9D%90%A8%F0%9D%90%AE%F0%9D%90%AB-30-%F0%9D%90%9D%F0%9D%90%9A%F0%9D%90%B2-%F0%9D%90%8C%F0%9D%90%8B%F0%9D%90%83%F0%9D%90%8B-activity-7407085094924042240-GuEn/
Day 30	Celebration & Reflection	A meaningful wrap-up of the 30-day journey — growth, lessons, mindset, and what comes next	Write a reflective README + record a final video to celebrate the journey	Celebrate the journey & mark the beginning of what's next	https://www.linkedin.com/posts/serhii-kravchenko1_%F0%9D%90%83%F0%9D%90%9A%F0%9D%90%B2-30-%F0%9D%90%8E%F0%9D%90%9F%F0%9D%90%9F%F0%9D%90%A2%F0%9D%90%9C%F0%9D%90%A2%F0%9D%90%9A%F0%9D%90%A5%F0%9D%90%A5%F0%9D%90%B2-%F0%9D%90%8E%F0%9D%90%AE%F0%9D%90%AB-activity-7410329423587049472-3FAk/

---

🧩 How to Use This Repository

1. Clone the repository:

git clone https://github.com/Serhii2009/ml_fundamentals_challenge
cd ml_fundamentals_challenge

2. Go through each folder day by day:

cd day_01_loss_function
# Open Python script and study it
python loss_function.py

3. Read the README.md in each folder for explanations, analogies, and exercises.

4. Practice by modifying the code, experimenting with parameters, and solving exercises.

5. Explain each topic in your own words (even to a 10-year-old!)—this is a key step for deep understanding.

---

💡 Notes

• All code is written in Python 3, using NumPy, pandas, scikit-learn, and PyTorch/Keras for deep learning examples.
• Each day builds on the previous, so it's recommended to follow the sequence from Day 1 to Day 30.
• This repository is designed for self-learning, teaching, and collaboration.

---

🎓 Outcome

By completing this 30-day challenge:

• You will have a solid understanding of ML and DL fundamentals.
• You will be able to implement algorithms from scratch and understand their inner workings.
• You will gain confidence to explain concepts clearly to others.
• You will have a structured portfolio of practical ML/DL projects.

Learning by doing, reflecting, and teaching is the fastest path to mastering Machine Learning and Deep Learning.

If you follow this repository day by day, and truly practice each topic, you will understand the math, the code, and the intuition behind every core concept.

---

📝 License

This repository is licensed under the MIT License – see the LICENSE file for details.