Data Science + AI/ML

• Python setup
• VS Code
• Jupyter Notebook
• Variables
• Data types
• Strings
• Lists
• Tuples
• Dictionaries
• Sets
• Loops
• Functions
• File handling
• CSV and JSON
• pip install
• Virtual environments
• List comprehension

• What is NumPy
• ndarray
• Creating arrays
• Shape and reshape
• Indexing and slicing
• Broadcasting
• Mathematical operations
• Statistical functions (mean median std)
• Random module
• Array performance vs lists
• Matrix operations basics

• Series and DataFrame
• read_csv read_excel
• head() tail() info() describe()
• Selecting columns rows
• Filtering (boolean indexing)
• Sorting
• Groupby and aggregation
• Merge join concat
• Pivot tables
• Handling missing values (fillna dropna)
• Data type conversion
• apply() and lambda
• String operations in Pandas
• Date and time handling

• Mean Median Mode
• Variance and Standard Deviation
• Percentiles and Quartiles
• Normal distribution
• Skewness and Kurtosis
• Probability basics
• Bayes theorem (intuition)
• Correlation (Pearson Spearman)
• Hypothesis testing (t-test chi-square)
• p-value and significance
• Central Limit Theorem
• All calculations done in Python

• Matplotlib basics (line bar scatter hist)
• Subplots and figure size
• Labels titles legends
• Seaborn (countplot barplot boxplot heatmap violinplot pairplot)
• Customising colour palettes
• Visualisation best practices
• Telling a story with charts
• EDA visualisation workflow

• Choosing a real dataset (Kaggle)
• Data loading and inspection
• Cleaning and preprocessing
• Univariate and bivariate analysis
• Correlation heatmap
• Outlier detection (IQR Z-score)
• Feature insights and summary
• EDA report writing
• Presenting findings

• What is Machine Learning
• Supervised vs Unsupervised
• Train test split
• Linear Regression (theory + code)
• Multiple Linear Regression
• Polynomial Regression
• Evaluation metrics (MAE MSE RMSE R2)
• Feature scaling (StandardScaler MinMaxScaler)
• Handling categorical variables (LabelEncoder OneHotEncoder)
• Overfitting and underfitting
• Project: House Price Prediction

• Logistic Regression
• Decision Tree Classifier
• Random Forest Classifier
• K-Nearest Neighbours (KNN)
• Support Vector Machine (SVM)
• Evaluation metrics (accuracy precision recall F1 confusion matrix ROC-AUC)
• Cross-validation
• Hyperparameter tuning (GridSearchCV)
• Project: Customer Churn Prediction
• Project: Loan Approval Classifier

• K-Means Clustering
• Elbow method
• Silhouette score
• PCA (Principal Component Analysis)
• Dimensionality reduction
• Feature engineering techniques
• Feature selection (correlation variance importance)
• Handling imbalanced datasets (SMOTE overview)
• Pipeline concept in Scikit-learn

• What are Neural Networks
• Perceptron concept
• Activation functions (ReLU Sigmoid Softmax)
• TensorFlow and Keras setup
• Building a simple neural network
• Training and evaluation
• Image classification (MNIST handwritten digits)
• NLP basics (text preprocessing tokenisation)
• Sentiment analysis with simple model
• GPU vs CPU (awareness)
• What to learn next in deep learning

• Git and GitHub for data projects
• Jupyter Notebook best practices
• Saving and loading models (pickle joblib)
• Streamlit basics (build a simple ML web app)
• Deploying a Streamlit app (Streamlit Cloud)
• Kaggle account and competitions overview
• Google Colab for heavy computation

• Capstone project (end-to-end ML pipeline)
• Dataset selection
• EDA and preprocessing
• Model building and comparison
• Model evaluation and selection
• Documentation and README
• GitHub upload
• Streamlit demo app
• Resume building for data roles
• LinkedIn optimisation
• Portfolio presentation
• Mock interviews (technical + HR)
• Freelance data analysis guide
• Career roadmap (analyst to scientist to engineer)

Perform a complete EDA on a real e-commerce transactions dataset. Load the data with Pandas, handle missing values, analyse sales trends by category and region, create 10+ visualisations with Matplotlib and Seaborn, and write a summary report with actionable business insights.

Build a Linear Regression model that predicts house prices based on features like area, number of bedrooms, location, and age. Use the Ames Housing dataset from Kaggle. Handle missing values, encode categorical features, scale numerical features, evaluate with R2 and RMSE, and visualise predicted vs actual prices.

Build a classification model (Random Forest) that predicts whether a telecom customer will leave the company. Feature engineering, handling class imbalance, cross-validation, hyperparameter tuning with GridSearchCV. Evaluate with confusion matrix, precision, recall, and F1-score.

Build a basic collaborative filtering recommendation system using the MovieLens dataset. Calculate similarity between users or items, generate top-N recommendations, and evaluate with simple metrics. Understand how Netflix and Amazon recommendations work at a conceptual level.

Build a simple NLP pipeline that classifies product reviews as positive or negative. Text preprocessing (lowercase, stopwords, tokenisation), TF-IDF vectorisation, train a Logistic Regression or Naive Bayes classifier, and evaluate accuracy. Understand the basics of how opinion mining works.

Take one of your trained models (house price or churn) and build a simple web interface using Streamlit. Users enter feature values (area, bedrooms, income) and the app shows the prediction in real time. Deploy the app on Streamlit Cloud so anyone can access it via a URL.

Choose a real-world problem and dataset of your choice. Perform complete EDA, clean and preprocess the data, try multiple ML algorithms, compare model performance, select the best model, document your findings, deploy as a Streamlit app, and upload everything to GitHub with a professional README. This is your flagship portfolio project — the one you show in every interview.