Interview Prep

Question 1

Bias-Variance Tradeoff

Answer 1

Bias is error due to erroneous or overly simplistic assumptions in algorithm. This can lead to the model underfitting your data, making it hard to generalize your knowledge from the training set to the test set.

Variance is error due to too much complexity in algorithm. Leads to the algorithm being highly sensitive and overfitting.

Question 2

Supervised vs unsupervised learning

Answer 2

Supervised = used data that is labeled

Unsupervised = not labeled

Question 3

KNN vs k-means clustering

Answer 3

KNN = supervised classification algorithm

K-means clustering is unsupervised

Question 4

ROC Curve

Answer 4

Receiver operating characteristic –> representation of true positive rates vs false positive rates

Question 5

Precision vs. recall vs. accuracy

Answer 5

Email spam

Precision = TP / predicted positive (actual cancer patients / pred cancer patients)

Recall = TP / Real positive (Actual cancer patients / actual cancer patients + cancer patients not predicated)

Question 6

Type I vs Type II

Answer 6

Type I = False positive

Type II = False negative

Question 7

Cross-validation

Answer 7

Holding out different parts of the data to test the model on

Question 8

Decision tree pruning

Answer 8

Branches that have lower predictive power are removed in order to reduce complexity

Question 9

F1 score

Answer 9

weighted avg between recall and precision

1 = best
0 = worst

Question 10

How to avoid overfitting

Answer 10

Keep models simple

Use regularization (Lasso / Ridge)

Use cross-validation

Question 11

Examples of ensemble

Answer 11

Decision Tree + Boosting

Question 12

How would you handle missing or incomplete data?

Answer 12

Delete row or decide to replace with 0 or another value (mean or median) or prediction

Question 13

Write pseudocode for linear regression

Answer 13

1) Define cost function (total squared error) to minimize
2) Initialize gradient descent
3) Iterate on gradient decent based on alpha (learning rate) and number of iterations
4) Analyze results

Question 14

Why and how to normalize data

Answer 14

Ensure all data is on the same scale

= x - mean / SD

Question 15

Data analysis process

Answer 15

1) Get access to data
2) EDA to get familiar with data + understand any issues
3) Determine how to prepare the data
4) Explore models to use
5) Execute, visualize, and inform

Question 16

Difference between variance and covariance

Answer 16

Variance = how apart numbers are from mean

Covariance = how the mean of two variables move together

Question 17

Elements of CLT

Answer 17

with a large enough sample size, the sample mean and sd will match the population – you can use z-score to get prob

Question 18

Conditions for linear regression

Answer 18

1. There is a linear relationship between the dependent variables and the regressors
2. The errors or residuals of the data are normally distributed and independent from each other
3. There is minimal multicollinearity between explanatory variables
4. Homoscedasticity. This means the variance around the regression line is the same for all values of the

Question 19

Python data structures and which are mutable?

Answer 19

Lists, Dictionaries, Sets, Strings, Tuples, Boolean, Numeric

Tuples are immutable (along with frozensets)

Question 20

How best to split up test vs train

Answer 20

Use train, test, and validation – validation allows you to fine tune parameters before running final test

80:20 typically appropriate but depends on the data you have

Question 21

Regression algorithms

Answer 21

Linear regression
Regression trees
Neural Network

Question 22

Type of classification algorithms

Answer 22

Support Vector Machines
Decisions Trees
KNN

Question 23

KNN vs SVM

Answer 23

KNN calculates similarity scores

SMV - looks at difference between hyperplanes

Question 24

Git commands

Answer 24

git pull
git add file/folder
git commit -m “comment”
git push

Question 25

explain neural networks

Answer 25

Replicates now the brain works

Input layer: Info comes in to an input layer
Hidden layer: n # of hidden layers, that make computations to make sense of pieces of info
Output layer: main outcome

Question 26

graph database

Answer 26

neo4j = graph db, technically non-SQL, user cypher language
neptune (AWS) using SPARQL, gremlin
stores in shape of graph/network (nodes and edges)