PopularMajor Project
Machine Learning
AI Resume Screening System
NLP pipeline that ranks candidates against a job description
- Python
- spaCy
- scikit-learn
- FastAPI
- +2
₹4,499₹7,999
4.9/214 reviews
Machine Learning · Major Project
Click-Through Rate Prediction
Predicts ad clicks with feature hashing and gradient boosting
3-4 weeks build
Overview
Works on a sample of the public Avazu click-through dataset from Kaggle, where most features are high-cardinality categoricals. The kit compares one-hot encoding, feature hashing and target encoding, trains logistic regression and LightGBM, and evaluates with log loss and ROC-AUC on a time-ordered split. A FastAPI endpoint returns a click probability for an ad impression.
What makes it stand up
- 01Handles millions of rows with chunked loading and sampling
- 02Feature hashing compared with one-hot and target encoding
- 03Logistic regression baseline against LightGBM
- 04Log loss, ROC-AUC and calibration on a time-ordered split
- 05FastAPI scoring endpoint for single impressions
Module breakdown
- 01
Data handling
Chunked reads, sampling and memory-efficient dtypes.
- 02
Encoding
Hashing trick, one-hot and target encoding pipelines.
- 03
Models
Logistic regression and LightGBM with early stopping.
- 04
Evaluation
Log loss, AUC and reliability diagrams.
- 05
Scoring API
FastAPI service returning a click probability.
After this, you will be able to
- Explain the hashing trick and the collisions it causes
- Defend log loss as the metric for probability estimates
- Demo a live prediction for a sample impression
Technology stack
- Python
- Pandas
- scikit-learn
- LightGBM
- FastAPI
- Level
- Major Project
- Domain
- Machine Learning
- Build time
- 3-4 weeks
- Difficulty
You receive
- Complete, commented source code
- 60-90 page project report (IEEE format)
- Editable presentation deck
- Architecture, ER and UML diagrams
- Local setup and deployment guide
- Sample dataset or seed data
- Viva question bank with answers
- One 45-minute walkthrough call
- 30 days of doubt-clearing support