Projects completed as a part of IIIT-Delhi's Post Graduation Diploma in Artificial Intelligence.
$ git clone https://github.com/nanditashukla/AIML-Projects.git
$ cd AIML-Projects
1. Supervised Machine Learning
- Skills & Tools Covered: Exploratory Data Analysis, Data Preprocessing, Logistic regression, Naive Bayes classifiers, KNN classification
- Project link: Supervised Machine Learning
- Project Topic: Telecom Customer Churn Prediction
- Built Supervised Learning Classification models to help identify potential customers who have a higher probability to churn. Compared models built with Logistic Regression, KNN algorithm and Naive Bayes in order to select the best performing one.
2. Unsupervised Machine Learning
- Skills & Tools Covered: K-means clustering, Agglomerative Hierarchical clustering, Dimension Reduction-PCA
- Project link: Unsupervised Learning
- Project Topic: Mobile User Demographics - clustering
- Built an Unsupervised Learning model by making use of clustering algorithms to cluster user segments with similar interests/habits based on silhouettes and used PCA in order to reduce dimensionality, to help the App/mobile providers better understand and interact with their subscribers.
3. Ensemble Methods
- Skills & Tools Covered: Decision Trees, Bagging, Random Forests, Boosting
- Project link: Ensemble Methods
- Project Topic: Google Store App Rating Prediction
- Leveraged app data and user ratings from the app stores and extracted insightful information to predict rating for a given app. Decision tree classification algorithm was used and ensemble techniques like Random forest, Bagging, Gradient boosting, ADA boosting and Stacking were used to further improve the classification results.
4. Feature Engineering Techniques
- Skills & Tools Covered: Exploratory Data Analysis, Feature Exploration and Selection Techniques, Hyperparameter Tuning, Handling imbalanced data
- Project link: Feature Engineering Techniques
- Project Topic: Bank Loan Defaulter Prediction
- Used feature exploration and selection technique to predict if a customer will be a loan defaulter or not based on the given input features such as funded amount, term, interest rate etc. Used ensemble method Random forest to find out the most important features. Cross-validation techniques and grid search were used to tune the parameters for the best model performance.
5. Neural Networks
- Skills & Tools Covered: Batch Normalization, Hyper parameter tuning, Tensor Flow & Keras for Neural Networks & Deep Learning
- Project link: Neural Networks
- Project Topic: Signal Quality prediction of communication equipment using Neural Networks
- Build a deep learning model which can help the communications company to predict the equipment’s signal quality using various parameters. Used batch normalization, kernel initializer and dropout layers to fine tune the model and obtained an improvement in performance metrics.