Sangamesh Chandankera Email & Phone Number

Indore, Madhya Pradesh, India

Bengaluru, Karnataka, India

Tags:- Deep Learning, Computer Vision, Image Preprocessing, Transfer Learning, Model DeploymentBuilt an app that detects the sentiment of the online classroom using live video from the webcam and real-time aggregated feedback to the instructors.Employed essential image preprocessing techniques such as image augmentation, Pixel brightness transformations, etc. to improve image quality.Deployed quantized model on Streamlit API that generates sentiment of attending students aggregates… Show more Tags:- Deep Learning, Computer Vision, Image Preprocessing, Transfer Learning, Model DeploymentBuilt an app that detects the sentiment of the online classroom using live video from the webcam and real-time aggregated feedback to the instructors.Employed essential image preprocessing techniques such as image augmentation, Pixel brightness transformations, etc. to improve image quality.Deployed quantized model on Streamlit API that generates sentiment of attending students aggregates the classroom overall sentiment.Experimented with multiple pre-trained models such as multi-task CNN , VGG16, and MobileNetV2 to reduce the training time and improved accuracy, the final model built with a class average accuracy of 82%. Show less

Tags:- k means, PCA, Silhouette and Elbow scoreOnline Retail Customer Segmentation(07/2021-02/2021)Built a clustering model using k-means clustering and PCA on a transactional dataset of a UK based and registered non store online retail.Applied feature engineering to obtain new features such as Recency , Frequency , Monetary and RFMscore for getting more details about the customer.Obtain the optimal number of clusters using silhouette analysis and Elbow method and Similarlity… Show more Tags:- k means, PCA, Silhouette and Elbow scoreOnline Retail Customer Segmentation(07/2021-02/2021)Built a clustering model using k-means clustering and PCA on a transactional dataset of a UK based and registered non store online retail.Applied feature engineering to obtain new features such as Recency , Frequency , Monetary and RFMscore for getting more details about the customer.Obtain the optimal number of clusters using silhouette analysis and Elbow method and Similarlity identifed the optimal clusters and dendrogram from hierarchical clustering.Generated optimal customer clusters which could be further used in various use-cases like customer targeting, marketing campaigns, churn prediction,etc. Show less

Tags : ClassificationTASK : Your task is to build a model that predicts price range indicating how high the price is. In the competitive mobile phone market companies want to understand sales dataof mobile phones and factors which drive the prices.The mobile phone market is competitive, to say the least. In times where a mid-range phone offers the same if not more features, we still see brands' high-end phones dominating the market share.The most important features… Show more Tags : ClassificationTASK : Your task is to build a model that predicts price range indicating how high the price is. In the competitive mobile phone market companies want to understand sales dataof mobile phones and factors which drive the prices.The mobile phone market is competitive, to say the least. In times where a mid-range phone offers the same if not more features, we still see brands' high-end phones dominating the market share.The most important features in predicting a mobile phone's price are ram, battery power, pixel height and pixel width! We figured this out by using a correlation matrix, specifically looking at the most highly correlated variables to price range.Used machine learning to predict mobile phones price's using all of the features in our dataset. Show less

Achievements/Tasks:-Built a regression model using GBM , Decision tree regressor , LightGBM and XGBoost models to predict taxi trip time in NYC for a time period of two months.Used folium graph to visualize the pick-up and drop-off locations and used heatmaps to check for outliers and correlation of features with each other.Employed pre-processing techniques such as feature scaling, outlier treatment, missing value imputation and performed temporal sampling to generate train and… Show more Achievements/Tasks:-Built a regression model using GBM , Decision tree regressor , LightGBM and XGBoost models to predict taxi trip time in NYC for a time period of two months.Used folium graph to visualize the pick-up and drop-off locations and used heatmaps to check for outliers and correlation of features with each other.Employed pre-processing techniques such as feature scaling, outlier treatment, missing value imputation and performed temporal sampling to generate train and test data.LightGBM is more efficient than other models with an acurracy of 72% to predict the trip duration and help them to select suitable route . Show less