Swiggy

"Round 1 Overview - Interviewer will ask everything possible which is written on your resume • Started with my introduction. Asked me if I have a favourite project or should he choose any. I had a favourite one , From where did you get data, a few questions on Data preprocessing ,Correlation formula & function, What does a heat-map do? How did you handle class imbalance? Ideal data size for a ML Model? Which is more prone to Overfitting Decision Tree or RandomForest and why? In this case why DecisionTreeClassifier has more accuracy than RandomForest? What is OOB Score? Mathematical Derivation of how OOB Score is 37%. On which features did you use Log Transform and why? Switched to a project on NLP. Problem Statement? Use Case/Motivation? Derive Formula for TFIDF Vectorizer and full form. What is Bag of words? What is Keras and what is Tensorflow? Why did you choose randomforest instead of Logistic Regresion. Are there any accuracy metrics other than Accuracy score? Precision , Recall formula and what should be high Precision and Recall , why ?for a healthcare based classification project. Loss function of Logistic Regression , Formula of probability in logistic regression. Round 2 Overview - ML Coding ( Python , Numpy , Pandas ) & SQL , No DSA was asked. ML Coding - k=5 ( no. of clusters ) 50 dimensional data 1000 data points randomly define cluster centroids , data map to cluster centroid , create data. SQL - Table contains Employee_name and Salary , return employee_name who has 5th highest salary"