I worked towards solving data science-related questions and queries over the doubt resolution forum.Gained proficiency in tools and technologies like Python, SQL, sk learn and teamwork, time management & problem-solving skills.

* Developed an app using multiple CNN architectures that detects the sentiment of theonline classroom using live video from the webcam and give feedback to tutor.* Employed essential image preprocessing techniques such as image augmentation,Pixel brightness transformations etc. Use RELU, Dropout, Max-pool etc.* Experimented CNN such as LeNet, AlexNet, VGG16, VGG19 etc. Also used transferlearning to improve model accuracy. The model achieved 67% accuracy on test data.

* Built several clustering models using K-Means, Hierarchical and DBSCAN Clustering toidentify major customer segments on transactional Dataset of an online retail.* Performed EDA and RFM modelling by feature engineering to obtained new featuressuch as Recency, Frequency, Monetary, RFM, RFMScores etc.* Obtained the optimal number of clusters using silhouette analysis, Elbow method anddendrogram from hierarchical clustering. Use DBSCAN to deal with outliers.

* Developed classification model to predict ten-year risk of weather a person will sufferfrom Coronary Disease by trying Machine Learning classification algorithm* Performed EDA to obtain meaningful insights. Did Feature Engineering, experimentedalgorithms such as Logistic regression, Naïve Bayes, SVM, Random Forest classifiers.* Tuned the hyperparameters to get optimal model performance and evaluated themodels support vector classifier gave the best Recall of 71%.

* Developed a supervised regression based model and predict the hourly demand of bikes to reduced the public waiting time using Machine Learning Algorithms* Performed data cleaning, outlier handling, stastical analysis, EDA to gain meaningful insights and Feature Engineering to extract features for the best model performance.* The model was experimented using Linear Regression but didn't perform well so, finally using Random Forest and Gradient Boost we achieved Adjusted R2 of 0.91