A structured learning-based repository containing a step-by-step, end-to-end walkthrough of the machine learning lifecycle—from Python fundamentals and data preprocessing to model training and deployment.
-
Python Foundations
- Core concepts such as
Python basics, fundamental data types, loops, functions, and error handling.
- Core concepts such as
-
Data Manipulation & Visualization
-
In-depth use of libraries including:
- Pandas: Series, DataFrame operations, filtering, grouping, merging.
- Matplotlib & Seaborn: Plotting essentials—line charts, histograms, scatter plots, heatmaps, pairplots.
-
-
Complete ML Workflow
- Data Extraction & Loading: Reading CSVs, Excel, SQL connectors.
- Exploratory Data Analysis (EDA): Statistical summaries, distribution analysis, correlations, visualizations.
- Data Cleaning: Handling missing data, duplicates, outliers, and data normalization.
- Feature Engineering & Preprocessing: Encoding categorical variables, scaling, and transforming features.
- Model Development: Training models like Logistic Regression, Decision Trees, Naive Bayes; evaluating their performance.
- Model Persistence: Saving trained models and vectorizers with
picklefor reuse.
-
Data_Cleaning Implements critical data hygiene routines—null value imputation, deduplication, and error handling.
-
EDA Visualizes dataset features, distributions, correlations, and general statistics for better insight.
-
Data_Preprocessing Prepares data for modeling: encoding, scaling, vectorizing, and processing text features if any.
-
Model_Building Trains models, evaluates on validation/test set using metrics like accuracy, precision, recall, and saves final models (
model.pkl,vectorizer.pkl) for deployment.
- Comprehensive Learning Guide: Documents every step of progressing from Python basics to end-to-end ML model deployment.
- Progression in Abstraction: Gradually builds from raw data handling to sophisticated modeling workflows.
- Reusable Components: Each script serves as a standalone piece—adaptable, modular, and easy to understand.
An interactive machine learning web application that detects whether a given text message (email or SMS) is Spam or Not Spam (Ham). Built using Python, trained on real-world labeled data, and deployed locally via Streamlit.
🎯 Goal: Empower users to test spam detection in real-time using natural language processing and machine learning.
📦 Email-SMS-Spam-Classifier/
├── 📁 models2/
│ ├── app2.py # Streamlit UI logic and model inference
│ ├── model.pkl # Trained ML model (pickle)
│ └── vectorizer.pkl # Text vectorizer (TF-IDF or CountVectorizer)
│
├── 📁 ProjectFile/
│ └── 📁 models/
│ ├── 1.Data_Cleaning.py
│ ├── 2.EDA.py
│ ├── 3. Data Preprocessing.py
│ └── 4. Model Building.py
│ ├── model.pkl
│ └── vectorizer.pkl
│
├── 📁 Project Graphs/
│ ├── Correlation.png
│ ├── Distribution Plot.png
│ ├── Histogram.png
│ ├── Word Cloud.png
│ ├── Top 30 Words.png
│ └── Model Comparison.png
| Tool | Purpose |
|---|---|
| Python | Programming Language |
| Pandas | Data manipulation |
| Scikit-learn | ML Algorithms + Metrics |
| Matplotlib / Seaborn | Data Visualization |
| Streamlit | Frontend UI for Predictions |
| Pickle | Model and vectorizer export |
git clone https://github.com/KshitijSawant1/Machine-Learning---Domain-Internship.git
cd Machine-Learning---Domain-Internshipcd ProjectFile/models
# Run each script sequentially:
python 1.Data_Cleaning.py
python 2.EDA.py
python 3.Data_Preprocessing.py
python 4.Model_Building.py
cd ../../models2
streamlit run app2.py
✅ Hands-on experience with:
- Real-world spam datasets
- NLP and feature extraction
- Model performance comparison
- Web UI development using Streamlit
- Model serialization and deployment pipeline
- Deploy on public server (e.g., HuggingFace Spaces / Heroku)
- Add LSTM or BERT-based spam detection
- Save message prediction history
- Support multiple languages
├── ProjectFile/
│ └── models/
│ ├── 1.Data\_Cleaning.py
│ ├── 2.EDA.py
│ ├── 3.Data\_Preprocessing.py
│ └── 4.Model\_Building.py
│
├── models2/
│ ├── app2.py # Streamlit-based UI for prediction demo
│ ├── model.pkl # Saved trained model
│ └── vectorizer.pkl # Saved feature vectorizer
│
├── Project Graphs/
│ ├── Correlation.png
│ ├── Distribution Plot.png
│ ├── Histogram.png
│ └── Word Cloud.png
│
└── README.md
- Fundamentals of Python programming: variables, loops, functions, error handling.
- Hands-on usage of
pandasfor data manipulation, andmatplotlib+seabornfor visual analytics.
1.Data_Cleaning.py: Handling missing values, outliers, duplicates.2.EDA.py: Exploring data distributions, correlations, and generating insight-focused visuals.
3.Data_Preprocessing.py: Encoding categorical data, scaling numerical features, and text vectorization.
4.Model_Building.py: Training machine learning models (e.g., logistic regression, Naive Bayes), evaluating performance, and serializing:model.pkl,vectorizer.pkl.
- Live predictive UI using Streamlit, located in
models2/app2.py. - Allows users to input text and view spam/ham classification results in real-time with the trained model.
Check out the Project Graphs/ folder for generated visual outputs:
- Correlation heatmaps
- Distribution histograms
- Word cloud visualizations, among others
- Python — Core language for scripting
pandas,numpy— Data handlingmatplotlib,seaborn— Visualizationscikit-learn— Model building and evaluation- Streamlit — Easy deployment of frontend UI
pickle— Model persistence
- Educational pipeline structure — Great reference for learning or teaching ML fundamentals.
- Modular and replicable — Each script tackles one stage of the ML workflow.
- Interactive demo experience — Immediate feedback via Streamlit UI.
Kshitij Sawant GitHub: @KshitijSawant1 Reach out via GitHub for feedback, suggestions, or project deployment assistance.