🧠 100 Days of Code: Machine Learning & AI with Python

Welcome to my #100DaysOfCode challenge focused on mastering Machine Learning and Artificial Intelligence using Python! 🚀

I’m learning in public, building projects, and sharing my journey every day. This repository contains my daily progress, notebooks, projects, and takeaways.

---

Follow along here 👉 Twitter: @mr_psychocyborg

---

📅 Challenge Duration

Start Date: [24-July-2025]
End Date: [ ]

---

✅ Daily Log

Day	Topic	Summary	Key Takeaway	Link
0	Introduction to Python	Installed Python & VSCode; ran first script	Python’s simplicity & readability make it beginner-friendly	Day-0
1	Python Basics	Variables, Data Types, and Operators	Python’s flexibility makes it great for beginners	Day-1
2	Control Flow	if/else statements, for and while loops, break/continue	Control flow is the backbone of logic building	Day-2
3	Functions, Modules, and File I/O	Defined reusable functions, imported built-in & custom modules, handled file read/write	Functions help break code into logical blocks; file I/O connects code with real-world data	Day-3
4	Python Data Structures	Learned Lists, Tuples, Dictionaries, and Sets	Each structure has unique strengths — choose the right one for the job	Day-4
5	CLI Calculator Challenge	Built a command-line calculator using Python functions	Practiced user input, conditionals & modular coding	Day-5
6	NumPy Basics	Learned about NumPy arrays, indexing, slicing, and broadcasting	NumPy provides fast, flexible, and powerful array operations	Day-6
7	NumPy - Math & Statistics	Learned NumPy’s built-in vectorized operations & statistical functions	NumPy makes math fast and easy using simple functions like sum(), mean(), std()	Day-7
8	Pandas - Series, DataFrames & Reading CSVs	Learned creating Series, DataFrames, filtering, reading CSVs	Pandas is great for working with tabular data and analyzing it easily	Day-8
9	Pandas - GroupBy, Merge, Pivot	Learned data aggregation, joining, and pivoting	Great for analyzing structured data like SQL but faster	Day-9
10	Matplotlib & Seaborn	Visualized data with plots, histograms, heatmaps	Clear charts = better data understanding & storytelling	Day-10
11	Exploratory Data Analysis (EDA) on Titanic Dataset	Visualized Titanic dataset using Matplotlib & Seaborn	Line, Bar, Histogram, Boxplot, Heatmap — data stories made clear	Day-11
12	EDA Continued + Handling Missing Values	Cleaned and filled missing data in Titanic dataset	Median, Mode, Drop columns — cleaner data, clearer insights	Day-12
13	SQL with Python – SELECT & JOIN (SQLite)	Queried mock Titanic-style data using SQLite & Pandas	SELECT, INNER JOIN, WHERE — combining SQL power in Python	Day-13
14	Project: Kaggle Dataset EDA (End-to-End)	Cleaned, explored, visualized a Kaggle dataset	SQL + Pandas + Seaborn — storytelling with real-world data	Day-14
15	Recap: Python for Data Science — Quiz + Notes	Revised DataFrames, filtering, aggregation concepts	Practiced quiz problems + noted key concepts for revision	Day-15
16	Intro to Machine Learning + Scikit-learn Pipeline	Built first ML workflow: preprocess → train → evaluate	Pipeline with StandardScaler + LogisticRegression on Titanic dataset	Day-16
17	Linear Regression Theory & Implementation	Learned LR concept, theory, and evaluation metrics (MSE, R²)	Implemented Linear Regression to predict house prices	Day-17
18	Linear Regression Project — House Prices	Data cleaning, feature selection, model training & evaluation	Predicted house prices using Linear Regression with selected features	Day-18
19	Logistic Regression + Classification Metrics	Trained model to classify Titanic passengers & evaluated performance	Classified survivors using Logistic Regression, evaluated with Accuracy, Precision, Recall, and F1-score	Day-19
20	Logistic Regression Mini Project (Titanic again!)	Improved preprocessing & feature engineering, re-trained Logistic Regression model	Classified Titanic survivors with higher accuracy & better balanced metrics	Day-20
21	Decision Trees – Intuition + Practice	Built & visualized Decision Tree classifier, explored splitting (Gini/Entropy), overfitting & pruning	Learned how trees form the basis for ensemble methods	Day-21
22	Random Forest – Bagging, Feature Importance	Implemented Random Forest using bagging, explored feature importance, and improved accuracy over single trees	Built a robust classifier with reduced overfitting and better performance	Day-22
23	Model Evaluation (Accuracy, Confusion Matrix)	Evaluated ML models beyond accuracy using precision, recall, F1-score	Improved understanding of balanced vs imbalanced metrics	Day-23
24	ROC, Precision-Recall, F1 Score	Evaluated models beyond accuracy using ROC-AUC, PR curves & F1 score	Learned to choose metrics for imbalanced datasets	Day-24
25	KNN + Model Selection (train_test_split)	Implemented KNN with dataset splitting, tuned neighbors (k)	Achieved balanced accuracy with proper validation	Day-25
26	SVMs (Linear & Non-linear)	Implemented SVM with linear & RBF kernels, compared decision boundaries	Observed how SVM handles separable vs. non-separable data	Day-26
27	Unsupervised ML: KMeans	Implemented KMeans clustering from scratch & with sklearn	Visualized clusters and understood centroid updates	Day-27
28	Clustering Project (Customer Segmentation)	Applied KMeans to segment customers using real-world data	Identified distinct customer groups and business insights	Day-28
29	Dimensionality Reduction (PCA, t-SNE)	Applied PCA & t-SNE on high-dimensional data to visualize clusters	Observed PCA's linear vs t-SNE's nonlinear separability	Day-29
30	Project: Spam vs Ham Classifier	Cleaned email text, applied TF-IDF, trained Naive Bayes	Classified messages as spam or ham with strong accuracy	Day-30
31	Deep Learning Intro: Perceptron, Backpropagation	Built single-layer perceptron from scratch using numpy	Learned how forward & backward propagation adjust weights	Day-31
32	TensorFlow/Keras Setup + First Neural Net	Installed TF/Keras, built first NN using Sequential API	Observed how weights update through training automatically	Day-32
33	Dense Neural Networks on MNIST	Implemented a fully connected neural net with ReLU + softmax using TensorFlow/Keras	Achieved >97% accuracy in digit classification	Day-33
34	Activation Functions & Optimizers	Tested ReLU, Sigmoid, Tanh & Softmax; compared SGD, Adam, RMSprop on DNN	Observed how activations & optimizers affect learning speed & accuracy	Day-34
35	CNNs (Convolutional Neural Nets)	Implemented CNN for image classification	Observed automatic feature extraction using convolution & pooling layers	Day-35
36	CNNs with CIFAR-10	Built and trained CNN with Conv2D, MaxPooling, Dropout on CIFAR-10	Observed improved accuracy with deeper layers & regularization	Day-36
37	Transfer Learning with pre-trained models (ResNet50)	Used ResNet50 pretrained on ImageNet, fine-tuned top layers	Achieved faster convergence & high accuracy with fewer data	Day-37
38	Image Classifier App	Built a local deployable app using an image dataset (e.g., flowers, fashion, pets)	Trained a lightweight classifier and integrated real-time predictions	Day-38
39	Deploying with Streamlit or Flask	Turned ML model into an interactive web app	Compared Streamlit’s ease vs Flask’s flexibility	Day-39
40	ML Concepts Recap	Created a comprehensive mind map of all concepts	Visualized interconnections between algorithms and techniques	Day-40
41	Intro to NLP + Text Preprocessing	Learned NLP basics, tokenization, stopword removal, stemming & lemmatization	Understood how text cleaning impacts downstream ML models	Day-41
42	Tokenization, Stopwords, Lemmatization	Implemented text preprocessing steps in NLP	Learned how tokenization & lemmatization improve feature quality	Day-42
43	Text Vectorization (BoW & TF-IDF)	Implemented Bag of Words & TF-IDF on sample text corpus	Learned how TF-IDF captures word importance beyond raw counts	Day-43
44	NLP Project: Movie Review Sentiment Analysis	Built a sentiment classifier using TF-IDF + Logistic Regression with a Streamlit UI	Created an interactive app to test sentiment predictions in real-time	Day-44
45	Word Embeddings (Word2Vec, GloVe)	Implemented Word2Vec on text corpus, explored semantic similarity & used pre-trained GloVe embeddings	Observed embeddings capture meaning (e.g., king - man + woman ≈ queen)	Day-45
46	Intro to Transformers & HuggingFace	Explored HuggingFace pipelines for sentiment analysis, zero-shot classification, and text generation	Learned how transformers use self-attention for powerful NLP tasks	Day-46
47	Using BERT for Classification	Fine-tuned a pre-trained BERT model for text classification using HuggingFace	Observed how transfer learning boosts NLP performance	Day-47
48	Fine-tuning Transformers on custom text	Fine-tuned DistilBERT on a labeled dataset (text → sentiment)	Observed transfer learning drastically reduces training time & improves accuracy	Day-48

📌 You’ll find the full list of all 100 days in the /Days/ folder.

---

🛠️ Tools & Libraries Used

• Python 3.x
• VS Code
• NumPy
• Pandas
• Matplotlib
• Seaborn
• Scikit-learn
• joblib
• re
• string
• tensorflow
• tourch
• tourchvision
• flask
• pillow
• streamlit
• nltk
• spacy
• streamlit
• gensim
• datasets
• transformers

---

📚 Books I'm Using

Here are some of the best books I’m referring to throughout this journey:

Book Title	Author(s)	Why I’m Using It
Code with Python	Suresh Sundaradasu & S Rama Shree	Best for Python fundamentals with hands-on exercises
Problem Solving and Python Programming	J Jayalakshmi, Dr. D Stalin Alex and B Mahesh Prabhu	Great for practical Python scripting
Hands-On Machine Learning with Scikit-Learn, Keras & TensorFlow	Aurélien Géron	My go-to for ML concepts and real-world examples
...

---

🧩 Projects & Capstones

Project Name	Description	Link
Calculator	Basic arithmetic operations	Open
Titanic SQL EDA	Exploratory Data Analysis using Titanic dataset from Kaggle with SQL-style insights	Open
House Prices Project	Predicting house prices using Linear Regression with feature selection and evaluation	Open
Logistic Regression Mini Project	Classifying Titanic passengers with Logistic Regression, evaluated using multiple metrics	Open
Customer Segmentation (KMeans)	Segmented customers by demographics & spending using KMeans clustering	Open
Spam vs Ham Classifier (NLP)	Built a text classifier using TF-IDF + Naive Bayes to classify SMS messages as spam or ham	Open
Image Classifier App (CNN + Flask)	Trained a CNN on flower dataset and deployed via Flask for real-time image classification	Open
NLP Project: Movie Review Sentiment Analysis	NLP project using TF-IDF + Logistic Regression, deployed with Streamlit frontend	Open
...	Coming soon

---

🧠 Handwritten Notes for ML/AI

Explore my handwritten notes as I learn Machine Learning and AI step by step during #100DaysOfCode.

📚 Notes Repository: Handwritten Notes for ML/AI

---

📝 Why Handwritten Notes?

“What we write by hand, we remember more clearly.”
These notes are my way of reinforcing core concepts and staying consistent. They’re sketched during learning sessions or after completing hands-on projects.

---

🤝 Let's Connect

• 🐦 Twitter: @mr_psychocyborg
• 🌐 Portfolio: https://rahul-dohare-portfolio.vercel.app/
• 📚 Other Hand Written Notes: Google Drive
• 💌 Mail: psychocyborg007@gmail.com.com

---

⭐️ Star This Repo

If you're also on a #100DaysOfCode journey or just love learning ML/AI, consider giving this repo a ⭐️!

---

“Learning by building in public is the fastest way to grow. Let’s do this together. 💪”

RAHUL DOHARE