Used product reviews to check the sentiment analysis of the customer to the product
Random Train - Test Split (70 - 30) is done on dataset, TF and TFIDF (Unigram and Bigram) on the dataset is done to create sparse matrix
Evaluated the data with cross validation on several model building techniques such as Decision Tree, Logistic regression and also in deep learning techniques such as LSTM, ANN (Artificial Neural Network) to find the best model out of it and finally evaluated using test data.
Calculated accuracy, precision, recall using confusion matrix and also ROC and AUC curve were calculated for True positive rate vs False positive rate trade off

Used the patient record data to create a model which predicts when a patient will get admitted again in hospital.
Developed a classification model for predicting the probability of hospital readmission.
This output was used by insurance providers and hospital accounts team.
Re-admission risk was captured by categorizing patient attributes into numerical, categorical, and text attributes
Data from more 150,000 hospital admissions was used. Logistic regression model was used as the machine learning technique
Unsupervised learning method K-means clustering is used for clustering the data
Dimensionality reduction using PCA (Principle component Analysis) is used to reduce the dimension of the dataset
ROC curve and AUC measures were calculated for True Positive Rate vs False positive rate trade-offs

Built models to predict the availability of Amazon spot instances Imported Amazon spot price data into python and done data cleansing and Exploratory analysis like visual analysis, correlation analysis and also multi co linearity analysis Scaled the data if the variables are not comparable and also built PCA model and transform the original data to new dimension space Created Model Building such as Simple Linear, Simple Non-Linear Model, Multiple Linear Model, Multiple Non-Linear Model, Variable Transformation Executed Model metrics Evaluation using Visualize the fit (possible for simple regression), Random Train - Test Split (70 - 30), Validating using R-squared as well as Adjusted R squared for multiple regression and also done Cross validation on multiple train test splits

Used the patient record data to create a model which predicts when a patient will get admitted again in hospital. Developed a classification model for predicting the probability of hospital readmission. This output was used by insurance providers and hospital accounts team. Re-admission risk was captured by categorizing patient attributes into numerical, categorical, and text attributes Data from more 150,000 hospital admissions was used. Logistic regression model was used as the machine learning technique Unsupervised learning method K-means clustering is used for clustering the data Dimensionality reduction using PCA (Principle component Analysis) is used to reduce the dimension of the dataset ROC curve and AUC measures were calculated for True Positive Rate vs False positive rate trade-offs