Data Science Interview | 50 Frequently Asked Questions
1-What are the different Stages of building a Model ?
Answers: Define Problem Statement – Data Collection & Consolidation- Data Cleaning – Variable Selection – Exploratory Data Analysis – Model Development – Model Validation – Model Implementation – Model Documentation - Model Monitoring
2-What are the different variable reduction techniques you know?
Ans- Information Value Criteria, Cluster Analysis, Principal Component Analysis
3- What is Exploratory Data Analysis? What are the Exploratory Data Analysis techniques you know?
Ans – Finding descriptive Statistics, Visualize data with graphs, plots etc is known as Exploratory Data Analysis. Univariate & Multivariate techniques, Histograms, Scatter plots, QQ plot, Quantile distribution, Extreme Value detection methods, Principal Component Analysis, Cluster analysis
8- How do you validate the input datasets?
- Double checking data source
- Appling common sense. For ex. Age of a consumer can not be 200
- Compare with previous year data
- Doing basic data exploration and visualization
- Remove them from data, WOE Transformation, Log transformation etc
Define Concordance? How is it different R-Square?
- Concordance is the probability that when you pick an event (1) and an non event (0) from the sample randomly, the probability score from the Logistic regression for 1 is more than that of 0.
R Square explain variation explained in dependent variable
49- What are the means to check model fitness of the Logistic regression model?
- (Please watch video on model fitness on my channel)
50- What is deviance in a Logistic Regression Model?