InterView Experience: Fidelity Investments: CV shortlisting Interview Process There were two interviewer Introduce yourself What project you did I told him about Sentimental Analysis and PCA Explain Your first project(Sentimental Analysis) I explain them Creates Unique word vector Computed Term frequency Computed IDF Computed TF-IDF Then now we have data and now we can apply machine learning algorithm to it. Then he asked me did u do data preprocessing like staming or lametisation and what it means. I told them yeah I did data preprocessing like removing stopword and lowering the cases of the words because the word with different cases cant have different sentiments. But I told them that I didnt applied steming and all and told then I dont know their meaning after thinking for a few second. Any other algorithm you applied? I read an algorithm like SVM, Logistic regression, Linear regression, K-means clustering algorithms. Explain SVM. SVM is lot more similar to concept of perceptron but the main difference is that in perceptron we try to find a hyperplane that classify the two classes. But in SVM we are finding that hyperplan which are classifying classes and also maximising the margin. Basically we are getting hyperplane with maximum margin. What function you are using as loss function in Logistic regression? I was not knowing the name. But I told them that we are multiplying the different actual and predicted label and summing them up. And our main goal is to minimise that loss function. Then he cross questioned me that if you predicted all class 0 then what your loss function will be according to your arguments. I was getting the loss function in this case to be 0 but I was sure that it is wrong so I told them that I am making any mistake and apolozied for that. Then they said me its ok. And moved to next question. What matrix you use for accuracy in logistic regression. Confusion matrix. What is that? Basically It is matrix . I-th and J-th cell tells you number of data point predicted ‘I’ label but actually its label is ‘J’ How it help in calculating accuracy of your model? Fumbled…… Basically we compute F1-score by precision and recall(which I was not knowing and also they didnt ask) What was your major in Btech? CSE If we offer you data analytical position will it be more Interesting to you. I am new to data scientist I will try to explore new position. But If get data scientist role then It would be much preferable.